CA1341395C - Preparation of functional human factor viii - Google Patents

Abstract

The full DNA coding sequence of human factor VIII is identified herein. Also disclosed is the recombinant means useful to isolate and express this coding sequence in the preparation of functional human factor VIII polypeptide and functional derivatives thereof. Particularly preferred embodiments provide proof of unequivocal identification and functionality as well as specific, operable recombinant expression vectors and cultures.
This invention marks the successful application of recombinant DNA technology in the preparation of the largest and most elusive of human proteins, a goal with technical problems probably insurmountable to workers of ordinary skill having mere possession, e.g., of a partial or full clone encoding factor VIII.

Description

~ 34~ 39 5 PREPARATION OF,F N ,f1 N/1 L_ ~IL.IMAN FACTOR VIII
'This application is a division of application Serial Number 479,409 sled April 20, 1985.
Field of the Invention The present invention relates t:o human factor VIII, to novel ~5 forms and compositions thereof and particularly to means and methods for the preparation of functional species of human factor VIII, particularly via recombinant DtdA tectrnology.
The present invention is based in part on the discovery of the DNA sequence and deduced amino <acid sequence of human factor VIII as 20 well as associated portions c>f t;he factor VIII molecule found in our hands to be functional bioact:ive moieties. This discovery was enabled by the production of factor VIII in various forms via the application of recombinant Dh;A tee~hnoiody, thus, in turn enabling the production of sufficient quality and quantity of materials with 25 which to conduct biological testing and prove biological function-ality. Having dcterminc:ed such, it is poi>vit>le to tailor-make func-tional species of factor~ VIII via gi~netic~ manipulation and in vitro processing, arriving effic_ic~nt:ly at hitherto unobtainable commercial-ly practical amounts of ac:tive factor VIII products. Thi > invention is directed to these associated erol~odin~fm°rrts in all respects.
The publications and other materials hereof used to illuminate the background of the irr~rention, are listed at the end of the specification in the form of' a bibliography.

1 34~ 39 5 Background of the Invention The maintenance of an intact vascular system requires the interaction of a variety of cells and proteins. Upon injury to the vascular bed, a series of reactions is initiated in order to prevent fluid loss. The initial response is the activation of platelets, which adhere to the wound and undergo a series of reactions. These reactions include the attraction of- other platelets to the site, the release of a number of organic compounds and proteins, and the formation of a thrombogenic surface for the activation of the blood coagulation cascade. Through this combined series of reactions, a platelet plug is formed sealing the wound. The platelet plug is stabilized by the formation of fibrin threads around the plug preventing unwanted fluid loss. The platelet plug and fibrin matrix are subsequently slowly dissolved as the wound is repaired. For a general review, see (1).
A critical factor in the arrest of bleeding is the activation of the coagulation cascade in order to stabilize the initial platelet plug. This system consists of over a dozen interacting proteins present in plasma as well as released and/or activated cellular proteins (2, 3). Each step in the cascade involves the activation of a specific inactive (zymogen) form of a protease to the catalytically active form. By international agreement (4), each protein of the cascade has been assigned a Roman numeral designation. The zymogen form of each is represented by the Roman numeral, while the activated corm is represented by the Roman numeral followed by a subscript "a". The activated farm of the protease at each step of the cascade catal.ytically activates the protease involved in the subsequent step in the cascade. In this manner a small initial stimulus resulting in the activation of a protein at the beginning of the cascade is catalytically amplified at each step such that the final outcome is the formation of a burst of thrombin, with the resulting thrombin catalyzed conversion of the soluble protein fibrinogen into its insoluble form, fibrin. Fibrin has the property of self-aggregating into threads or fibers which ~ 34~ 39 5 function to stabilize the platelet plug such that the plug is not easily dislodged.
The current understanding of the interactions of the proteins involved in blood coagulation is as follows. The lack or deficiency of any of the proteins involved in the cascade would result in a blockage of the propagation of the initial stimulus for the production of fibrin. In the middle of the cascade is a step wherein f:act.or LX;", in itiatE~s the conversion of factor X to the activated form, factor xa. Factor VIII (also synonomously referred to as factor VIIIC) is currently believed to function at this step, in the presence of phospholipid and calcium ions, as a cofactor; that is, it has no known function in itself, and is required to enhance the activity of factor IXa.
This step in the cascade is critical since the two most common hemophilia disorders have been determined to be caused by the decreased functioning of either factor VIII (hemophilia A or classic hemophilia) or-factor IXa (hemophilia B). Approximately 80 percent _ of hemophilia disorders are due to a deficiency of factor VIII. The clinical manifestation in both types of disorders are the same: a lack of sufficient fibrin formation required for platelet plug stabilization, resulting in a plug which is easily dislodged with subsequent rebleeding at the site. The relatively high frequency of factor VIII and factor IX deficiency when compared with the other factors in the coagulation cascade is due to their genetic linkage to the X-chromosome. A single defective allele of the gene for factor VIII or factor IX results in hemophilia in males, who have only one copy of the X chromosome. The other coagulation factors are autosomally linked and generally require the presence of two defective alleles to cause a blood coagulation disorder - a much less common event. Thus, hemophilia A and B are by far the most common hereditary blood clotting disorders and they occur nearly exclusively in males.
Several decades ago the mean age of death of hemophiliacs was 20 years or younger. Between the early I950's and the late 1960's, research into the factor VIII disorder ied to the treatment of _4_ 1 341 3g 5 hemophilia A initially with whole plasma and, later, with concentrates of factor VIII. The only source for human factor_YIII
has been human plasma. Gne factor contributing to the expense is the cost associated with obtaining large amounts of usable plasma.
Commercial firms must establish donation centers, reimburse donors, and maintain the plasma in a frozen state immediately after donation and through the shipment to the processing plant. The plasma samples are pooled into lots of over 1000 donors and processed. Due to the instability of the factor VIII activity, large losses are associated with the few simple purification procedures utilized to produce the concentrates (resulting in approximately a 15 percent recovery of activity). The resulting pharmaceutical products are highly impure, with a specific activity of 0.5 to 2 factor VIII
units per milligram of protein (one unit of factor YIIT activity is by definition the activity present in one milliliter of plasma).
The estimated purity of factor VIII concentrate is approximately 0.04 percent factor VIII protein by weight. This high impurity level is associated with a variety of serious complications including precipitated protein, hepatitis, and possibly the agent responsible for Acquired Immune Deficiency Syndrome. These disadvantages of the factor VIII concentrates are due to the instability of the plasma derived factor VIII, to its low level of purity, and to its derivation from a pool of multiple donors. This means that should one individual out of the thousand donors have, for example, hepatitis, the 4vhole lot would be tainted with the virus. Donors are screened for hepatitis B, but the concentrates are known to contain both hepatitis A and hepatitis non-A non-B.
Attempts to produce a product of higher purity result in unacceptably large losses in activity, thereby increasing the cost.
The history of purification of factor VIII illustrates the difficulty in working with this protein. This difficulty is due in large part to the instability and trace amounts of factor VIII
contained in whole blood. In the early 1g70's, a protein was characterized which was then believed to be factor VIII (5, 6, 7).
This protein was determined to be an aggregate of a subunit glycoprotein, the subunit demonstrating a molecular weight of approximately 240,000 daltons as determined by SDS gel electrophoresis. This subunit aggregated into a heterogeneous population of higher molecular weight species ranging from between one million and twenty million daltons. The protein was present in hemophiliac plasma, but missing in plasma of patients with von Willebrand',s disease, an autosomally transmitted genetic disorder characterized hw~ a prolonged kO eeding time and low levels of factor VIII (8). The theory then proposed was that this high molecular weight protein, termed von Willebrand factor (vWF) or factor VIII
related antigen (FVIIIRAg), was responsible for the coagulation defect in both diseases, with the protein being absent in von Willebrand's disease and somehow non-functional in classic hemophilia disease states (9). However, it was later observed that under certain conditions, notably high salt concentrations, the factor VIII activity could be separated from this protein believed responsible for the activity of factor VIII (10-20). Under these conditions, the factor YIII coagulant activity exhibited a molecular weight of 100,000 to 300,000. Since this time, great effort has concentrated on identifying and characterizing the proteins) responsible for the coagulant activity of factor VIII. However, the availability of but trace amounts of the protein in whole blood coupled with its instability have hampered such studies.
Efforts to isolate factor VIII proteins) from natural source, both human and animal, in varying states of purity, have been reported (21-27, 79). Because of the above mentioned problems, the possibility exists for the mistaken identification and subsequent cloning and expression of a contaminating protein in a factor VIII
preparation rather than the factor VIII protein intended. That this possibility is real is emphasized by the previously mentioned mistaken identification of von Willebrand protein as being the factor VIII coagulant protein. Confusion over the identification of factor VIII-like activity is also a distinct possibility. Either factor xa or thrombin would cause a shortening of the clotting time of various plasmas, including factor VIII deficient plasma, thereby 1 34~ 39 5 appearing to exhibit factor VIII-like activity unless the proper controls were performed. Certain cells are also known to produce activities which can function in a manner very similar to that expected of factor VIII (28, 29, 30). The latter reference (30) proves that this factor VIII-like activity is in fact a protein termed tissue factor. The same or similar material has also been purified from human placenta (31). This protein functions, in association with the plasma protein factor VII, at the same step as factor VIII and factor zxa, resulting in the activation of factor X
to factor xa-The burden of proof for expression of a recombinant factor VIII
would therefore rest on the proof of functional expression of what is unquestionably a factor VIII activity. Even were prior workers to show that they obtained a full or partial clone encoding all or a portion of factor VIII, the technical problems in the expression of a recombinant protein which is four times larger than any other recombinant protein expressed to date could well have proven insurmountable to workers of ordinary skill.
Summary of the Invention The potential artifacts and problems described above combine to suggest the need for close scrutiny of any claims of successful cloning and expression of human factor VIII. The success of the present invention is evidenced by:
1) Immunological cross-reactivity of antibodies raised against clone-derived factor VIII proteins with plasma-derived factor VIII proteins.
2) Cross-reaction of neutralizing monoclonal antibodies raised against human plasma factor VIII with protein encoded by the clone.
3) Identification of a genomic DNA corresponding to the factor VIII cDNA of the inventian as being located in the X-chromosome, where factor VIII gene is known to be encoded.
0457(_ 13+1395 _;T_ 4) Expression of a functional protein which exhibits:
a) Correction of factor VIII deficient plasma.
b) Activation of factor X to factor xa in the presence of factor zxa, calcium and phospholipid.
c) Inactivation of the activity observed in a) and b) by antibodies specific for factor VIII.
d) Binding of the activity to an immobilized monoclonal antibody column specific far factor VIII.
e) Activation of the factor VIII activity by thrombin.
f) Qinding of the activity to and subsequent elution from immobilized von Willebrand factor.
Thus, the present invention is based upon the successful use of recombinant DNA technology to produce functional human factor VIII, and in amounts sufficient to prove identification and functionality and to initiate and conduct animal and clinical testing as prerequisites to market approval. The praduct, human factor VIII, is suitable for use, in all of its functional forms, in the prophylactic or therapeutic treatment of human beings diagnosed to be deficient in factor VIII coagulant activity. Accordingly, the present invention, in one important aspect, is directed to methods of diagnosing and treating classic hemophilia (or hemophilia A) in human subjects using factor VIII and to suitable pharmaceutical compositions therefor.
The present invention further comprises essentially pure, functional human factor VIII. The product produced herein by genetically engineered appropriate host systems provides human factor VIII in therapeutically useful quantities and purities. In addition, the factor VIII hereof is free of the contaminants with which it is ordinarily associated in its non-recombinant cellular environment.
The present invention is also directed to DNA isolates as well as to DNA expression vehicles containing gene sequences encoding human factor VIII in expressible farm, to transformant host cell cultures thereof, capable of producing functional human factor YIII. In still further aspects, the present invention is directed to various processes useful far preparing said DNA isolates, DNA
expression vehicles, host cell cultures, and specific embodiments thereof. Still further, this invention i=_> directed to the preparation of fermentation cultures of said cell cultures.
Further, the present invention provides navel polypeptides comprising moiety(ies) corresponding to functional segments of human factor VIII. These novel polypeptides may represent the bioactive and/or antigenic determinant segments of native factor VIII. For example, such polypeptides are usef~ul for treating hemophiliacs per se, and particularly those who have developed neutralizing antibodies to factor VIII. In the latter instance, treatment of such patients with polypeptides bearing the requisite antigen determinants) could effectively bind such antibodies, thereby increasing the efficiency of treatment with palypeptides-bearing the bioactive portions of human factor VIII.
The factor VIII DNA isolates produced according to the present invention, encoding functional maiety(ies) of human factor VIII, find use in gene therapy, restoring factor° VIII activity in deficient subjects by incorporation of such DNA, for example, via hematopoetic stem cells.
Particularly Preferred Embodiment Human factor VIII is produced in functional form in a particularly suitable host cell system. This system comprises baby hamster kidney cells (BHK-21 (C-13), ATGC No. CCL 10) which have been transfected with an expression vector' comprising DNA encoding human factor YITI, including 3'- and 5'- untranslated DNA thereof and joined at the 3'- untranslated region with 3'- untranslated terminator DNA sequence, e.g., such as from hepatitis B surface antigen gene. Expression of the gene is driven by transcriptional and translational control elements contributed by the adenovirus v ~ _9_ 1 3 41 3 9 5 major late promoter together with its 5' spliced leader as well as elements derived from the SV40 replication origin region including transcriptional enhancer and promoter sequences. In addition, the expression vector may also contain a DHFR gene driven by an SV40 early promoter which confers gene amplification ability, and a selectable marker gene, e.g., neomycin resistance (which may be provided via cotransfection with a separate vector bearing neomycin resistance potential).
Description o_f the Drawings Figure 1. Diagrammatic rE~presentation of the current understanding of the inter acti.ans c.o: the proteins involved in blood coagulation as described can ~><~ge 3.
Figure 2. Melting of DNA in TMAC1 and 6x SSC. A: For each point ten duplicate aliquots of a DNA were first bound to nitrocellulose filters. These filters were then hybridized without formamide at 37~C as described in Methods. Pairs of spots were then washed in 6xSSC, 0.1 percent SDS (O) or 3.0 M TMACI, 50 mM Tris HC1, pH 8.0, 0.1 percent SDS, 2 mM EDTA (0) in 2'C increments from 38 to 56'C.
The melting temperature is the point where 50 percent of the hybridization intensity remained. B: A melting experiment as in panel A was performed by binding aliquots of pBR322 DNA to nitrocellulose filters. Probe fragments of various lengths were generated by digestion of pBR322 with Mspl, end-labeling of the fragments with 32P, and isolation on polyacrylamide gels. The probe fragments from 18 to 75 b were hybridized without formamide at 37~C and those from 46 to 1374 b in 40 percent formamide at 37~C as described in Methods. The filters were washed in 3.0 P1 tetramethylammonium chloride (TP1AC1), 50 mh1 Tris HC1, pH 8.0, 0.1 Percent SDS, 2 mM EDTA in 3AC increments to determine the melting temperature. (0) melting temperature determined for pBR322 l~ls~I
probe fragments, (e) melting temperatures in 3.0 M TMAC1 from panel A for 11-17 b probes.

.. 1 341395 Figure 3. Detection of the Factor YIII gene with probe 8.3. Left three panels: Southern blots of 46,XY (1X, male) DNA and 49,XXXXY
(4X) human DNA digested with EcoRI and BamHI were hybridized in 6xSSC, 50 mM sodium phosphate (pH 6.8), 5x Denhardt's solution, 0.1 g/1 boiled, sonicated salmon sperm DNA, 20 percent formamide at 42'C
as described in Methods. The ti~ree blots were washed in lxSSC, 0.1 percent SDS at the temperature indicated. Lane l, EcoRI 1X; lane 2, EcoRI 4X; lane 3, BamHI 1X; and lane 4, BamHT 4X. Lane M.
end-labeled aHindIII and X174 HaeIII digested marker fragments.
Right panel: One nitrocellulose filter from the a/4X library screen hybridized with probe 8.3. Arrows indicate two of the independent Factor VIII positive clones. Hybridization and washing for the library screen was as described above for the Southern blots, with a wash temperature of 37'C.
Figure 4. Map of the Human Factor VIII Cene.
The top line shows the positions and relative lengths of the 26 protein coding regions (Exons A to Z) in the Factor VIII gene. The direction of transcription is from left to right. The second line shows the scale of the map in kilobase pairs (kb). The location of the recognition sites for the 10 restriction enzymes that were used to map the Factor VIII gene are given in the next series of lines.
The open boxes represent the extent of human genomic DNA contained in each of the a phage (a114, x120, x222, x482, x599 and x605) and cosmid (p541, p542, p543, p612, 613, p624) clones. The bottom line shows the locations of probes used in the genomic screens and referred to in the text: 1)0.9 kb EcoRI/QamHI fragment from p543; 2) 2.4 kb EcoRI/BamHI fragment from x222; 3) 1.0 kb NdeI/BamHI triplet of fragments from x120; 4) oligonucieotide probe 8.3; 5) 2.5 kb StuI/EcoRI fragment from x114; 6) 1.1 kb EcoRI/BamHI fragment from x482; 7) 1.1 kb BamHI/EcoRI fragment from p542. Southern blot analysis of 46,XY and 49,XXXXY genomic DNA revealed no discernible differences in the organization of the Factor VIII gene.

Figure 5. Cosmid vector pGcos4. The 403 b annealed NincII fragment of ac1857S7 (Bethesda Research Lab.) containing the cas site was cloned in pBR322 from AvaI to PvuII to generate the plasmid pGcosl.
Separately, the 1624 b PvuII to Nael fragment of pFR400 (4 9n), containing an SV40 origin and promoter, a mutant dihydrofolate reductase gene, and hepatitis B surface antigen termination sequences, was cloned into the pBR322 AhaIII site to generate the plasmid mp33dhfr. A three-part legation and cloning was then performed with the 1497 b S~hT to NdeI fragment of pGcosl, the 3163 b NdeI to EcoRV fragment of mp33dhfr, and the 376 b EcoRY to S~hI
fragment of pKTl9 to generate the cosmid vector pGcos3. pKTl9 is a derivative of pBR322 in which the BamHI site in the tetracycline resistance gene has the mutated nitroguanosine treatment. pGcos4 was generated by cloning the synthetic 20mer, 5' AATTCGATCGGATCCGATCG, in the EcoRI sit.:e of pGcos3.
- Figure 6. Map of pESYDA. The 342 b PvuII-HindIII fragment of SV40 _ virus spanning the SV40 origin of replication and modified to be bounded by EcoRI sites (73), the polyadenylation site of hepatitis B
virus (HBY) surface antigen (49n), contained on a 580 by BamHI-Bc~II
fragment, and the pBR322 derivative pML (75) have been previously described. Between the EcoRI site following the SV40 early promoter and the BamHI site of HBV was inserted the PvuII-HindIII fragment (map coordinates 16.63-17.06 of Adenovirus 2) containing the donor splice site of the first late leader (position 16.65) immediately followed by the 84D by HindIII-Sacl fragment of Adenovirus 2 (position 7.67-9.97) (49j), containing the Elb acceptor splice site at map position 9.83. Between the donor and acceptor sites lie unique BgIII and HindIII sites far inserting genomic ONA fragments.
Figure 7. Analysis of RNA transcripts from pESYDA vectors.
Confluent 10 cm dishes of COS-7 cells (77) were transfected with 2 ug plasmid DNA using the modified OEAE-dextran method (84) as described (73). RNA was prepared 4 days post-transfection from cytoplasmic extracts (49n) and electrophoresed in denaturing 1 3~~ 39 5 formaldehyde-agarose gels. After transfer to nitrocellulose, filters were hybridized with the appropriate 32P-labelled DNA as described in Methods. Filters were washed in 2X SSC, 0.2 percent SDS at 42' and exposed to Kodak XR5 film. The position of the 28S
and 18S ribosomal RNAs are indicated by arrow in each panel.
The 9.4 kb BamHI fragment of aIl4 containing exon A (see Fig. 4) was cloned into the B~cl_II site of pESVDA (Fig. 6). Plasmid pESVDAlII.S contained the fragment inserted in the orientation such that the SY40 early promoter would transcribe the genomic fragment in the prayer (i.e., sense) direction. pESVDAl11.7 contains the 9.4 kb III fragment in the oppo~;it=e orientation. Plasmid pESVDA.S12.7 contains the 12.7 kb SacI fragment of ~1I4 inserted (by blunt end ligation) into the B~III site of pESYDA ire the same orientation as pESVDA1l1.6.
A. Hybridization of filters containing total cytoplasmic RNA from cells transfected Yrith pESYDA, pESYDAI11.7_and pESVDAl11.6.
pESVD RNA (lane 1), pESVDAl11.7 (lane 2), pESVDAl11.6 (lanes 3-5). Probed with Factor 8 exon A containing fragment (lanes 1-4) or 1800 b StuI/Bam fragment (lane 5). Faint cross-hybridization is seen to 18S RNA.
B. Hybridization of RNA with _StuI/BamHI probe ("intron probe").
RNA from: 1) pESVDA, polyA-; c') pESVDA, polyA+; 3) pESYDAl11.7, polyA-; 4) pESVDAl11.7, polyA+; 5) pESYDAl11.6, polyA-; 6) pESYDA111.6, polyA+. The small dark hybridizing band seen in lanes A5, B1, B3 and B5 probably represents hybridization to tRNA or to an Atu repeat sequence found in this region.
C. Comparison of cytoplasmic RNA fram pESYDAI11.6 (lane 1) and pESVDA.S127 (lane 2) probed with exon A containing fragment.
Note the slight size increase in lane 2 representing additional exon sequences contained in the larger genomic fragment.

Figure 8. Sequence of pESVDA.S127 cDNA clone S36. The DNA sequence of the human DNA insert is shown for the cDNA clone S36 obtained from the exon expression plasmid pESVDA.S127 (see infra for details). Vertical lines mark exon boundaries as determined by analysis of genomic and cDNA clones of factor vIII, and exons are lettered as in figure 4. Selected restriction endonuclease sites are indicated.
Figure 9. cDNA cloning. Factor VIII mRNA is depicted on the third line with the open bar representing the mature protein coding region, the hatched area the signal peptide coding region, and adjacent lines the untranslated regions of the message. The 5' end of the mRNA is at the left. Above this line is shown the extent of the exon B region of the genomic clone x222, and below the mRNA line are represented the six cDNA clanes from which were assembled the full length factor VIII clone (see text for details). cDNA
synthesis primers 1, 3_, 4 and oligo(dT) are shown with arrows depicting the direction of synthesis for which they primed.
Selected restriction endonuclease sites and a size scale in kilobases are included.
Figure 10. Sequence of Human Factor VIII Gene. The complete nucleotide sequence of the composite Factar VIII cDtdA clone is shown with nucleotides numbered at the left of each line. Number one represents the A of the translation initiation codon ATG. Negative numbers refer to 5' untranslated sequence. (mRNA mapping experiments suggest that Factor VIII mRNA extends approximately 60 nucleotides farther 5' than position -109 shown here.) The predicted protein sequence is shown above the DNA. Numbers above the amino acids are S1-19 for the predicted signal peptide, and 1-2332 for the predicted mature protein. "Op" denotes the opal translation stop codon TAG.
The 3' polyadenylation signal AATAAA is underlined and eight residues of the poly(A) tail (found in clone ac10.3) are shown. The sequence homologous to the synthetic oligonucleotide probe 8.3 has also been underlined (nucleotides 5557-5592). Selected restriction ~ 341 39 5 endonuclease cleavage sites are shown above the appropriate sequence. Nucleotides 2671-3217 represent sequence derived from genomic clones while the remainder represents cDNA sequence.
The complete DNA sequence of the protein coding region of the human factor VIII gene was also determined from the genomic clones we have described. Only two nucleotides differed from the sequence shown in this figure derived from cDNA clones (except for nucleotides 2671-3217). Nucleotide 3780 (underlined) is G in the cDNA clone, changing the amino acid codon 1241 from asp to glu.
Nucleotide 8728 (underlined) in the 3' untranslated region is A in the genomic clone.
Figure 11. Assembly of full length recombinant factor VIII
plasmid. See the text section 8a for details of the assembly of the plasmid pSVEFVIII containing the full length of human factor VIII
cDNA. The numbering of positionscliffers from those in the text and _ Figure 10 by 72bp. _ Figure 12. Assembly of the factor VIII expression plasmid. See the text section 8b for details of the assembly of the plasmid pAML3p.8c1 which directs the expression of functional human factor VIII in BHK cells.
Figure 13. Western Blot analysis of factor VIII using fusion protein antisera. Human factor VIII was separated an a 5-10 percent polyacrylamide gradient SDS gel according to the procedure of (81).
One lane of factor VIII was st;aine.i with silver (80). The remaining lanes of factor VIII were electrophoretically transferred to nitrocellulose for Western Blot analysis. Radiolabeled standards were applied into lanes adjacent to factor VIII in order to estimate the molecular weight of the observed bands. As indicated, the nitrocellulose strips were incubated with the appropriate antisera, washed, and probed with 125I protein A. The nitrocellulose sheets were subjected to autoradiography.

Figure 14. Analysis of fusion proteins using C8 monoclonal antibody. Fusion proteins 1, 3 and 4 were analyzed by Western blotting analysis for reactivity with the factor VIII specific monoclonal antibody C8.
Figure 15. Elution profile for high pressure liquid chromatography (HPLC) of factor VIII on a Toya Soda TSK 4000 SW column. The column was equilibrated and developed at room temperature with 0.1 percent SDS in 0.1 M sodium phosphate, pH 7Ø
Figure 16. Elution profile for reverse phase HPLC separation of factor VIII tryptic peptides. The separation was performed on a Synchropak RP-P C-18 column (0.46 cm x 25 cm, 10 microns) using a gradient elution of acetonitrile (1 percent to 70 percent in 200 minutes) in 0.1 percent trifluoroacetic acid. The arrow indicates the peak containing the peptide with the sequence AWAYFSDVDLEK.
Figure 17. Thrombin activation of purified factor VIII activity.
The cell supernatant was chromatographed on the C8 monoclonal resin, and dialyzed to remove elution buffer. 'Thrombin (25ng) was added at time 0. Aliquots were diluted 1:3 at the indicated times and assayed for coagulant activity. !!nits per ml were calculated from a standard curve of normal human plasma.
Detailed Description A. Definitions As used herein, "human factor VIII" denotes a functional Protein capable, in viva or in vitro, of correcting human factor VIII deficiencies, characterized, for example, by hemophilia A. The protein and associated activities are also referred to as factor YIIIC (FVIIIC) and factor VIII coagulant antigen (FVIIICAg)(31a). Such factor VIII is produced by recombinant cell culture systems in active forms) corresponding to 1 34~ 39 5 factor VIII activity native to human plasma. (One "unit" of human factor VIII activity has been defined as that activity present in one milliliter of normal human plasma.) The factor VIII protein produced herein is defined by means of determined DNA gene and amino acid sequencing, by physical characteristics and by biological activity.
Factor VIII has multiple degradation or processed forms in the natural state. These are proteolytically derived from a precursor, one chain protein, as demonstrated herein. The present invention provides such single chain protein and also provides for the production per se or via in vitro processing of a parent molecule of these various degradation products, and administration of these various degradation products, which have been shown also to be active. Such products contain functionally active portions) corresponding to native material.
Allelic variations likely exist. These variations may be demonstrated by one or more amino acid differences in the overall sequence or by deletions, substitutions, insertions or inversions of one or more amino acids in the overall sequence. In addition, the location of and degree of glycosylation may depend on the nature of the host cellular environment. Also, the potential exists, in the use of recombinant DNA technology, for the preparation of various human factor VIII derivatives, variously modified by resultant single or multiple amino acid deletions, substitutions, insertions or inversions, for example, by mE~anS Of site directed mutagenesis of the underlying DNA. In addition, fragments of human factor VIII, whether produced in vivo or in vitro, may possess requisite useful activity, as discussed above. All such allelic variations, glycosylated versions, modifications and fragments resulting in derivatives of factor VIII are included within the scope of this invention so long as they contain the functional segment of human factor VIII and the essential, characteristic human factor VIII
functional activity remains unaffected in kind. Such functional variants or modified derivatives are termed "human factor VIII

derivatives" herein. Those derivatives of factor YIII possessing the requisite functional activity can readily be identified by straightforward in vitro tests described herein. From the disclosure of the sequence of the human factor VIII DNA herein and the amino acid sequence of human factor VIII, the fragments that can be derived via restriction enzyme cutting of the DNA or proteolytic or other degradation of human factor VIII protein will be apparent to those skilled in the art.
Thus, human factor VIII in functional form, i.e., "functional human factor VIII", is capable of catalyzing the conversion of factor X to Xa in the presence of factor IXa, calcium, and phospholipid, as well as correcting the coagulation defect in plasma derived from hemophilia A affected individuals, and is further classified as °functional human factor VIII" based on immunological properties demonstrating identity or substantial identity with human plasma factor VIII.
_"Essentially pure form" when used to describe the state of "human factor VIII" produced by the invention means substantially free of protein or other materials ordinarily associated with factor VIII when isolated from non-recombinant sources, i.e. from its "native" plasma containing environment.
"DHFR protein" refers to a protein which is capable of exhibiting the activity associated with dihydrofolate reductase (DHFR) and which, therefore, is required to be produced by cells which are capable of survival on medium deficient in hypoxanthine, glycine, and thymidine (-HGT medium). In general, cells lacking DHFR protein are incapable of growing on this medium, and cells which contain DHFR protein are successful in doing so.
"Expression vector" includes vectors which are capable of expressing DNA sequences contained therein, where such sequences are operably linked to other sequences capable of effecting their expression. These expression vectors replicate in the host cell, either by means of an intact operable origin of replication or by functional integration into the cell chromosome. Again, "expression vector" is given a functional definition, and any DNA sequence which is capable of effecting expression of a specified ONA code disposed therein is included in this term as it is applied to the specified sequence. In general, expression vectors of utility in recombinant DNA techniques are often in the form of °°plasmids" which refer to circular double stranded DNA loops. However, the invention is intended to include such other forms of expression vectors which serve equivalent functions.
"D~dA isolate" means the ANA sequence comprising the sequence encoding human factor VIII, either itself or as incarporated into a cloning vector.
"Recombinant host cell" refers to cell/cells which have been transformed with vectors constructed using recombinant DNA
techniques. As defined herein, factor VIII or functional segments thereof are produced in the amounts achieved by virtue of this transforma~:ion, rather than in such lesser amounts, and degrees of purity, as might be produced by an untransformed, natural host source. Factor VIII produced by such "recombinant host cells" can be referred to as "recombinant human factor VIII".
Size units for DNA and RNA are often abbreviated as follows:
b~base or base pair; kb = kilo Eone thousand) base or kilobase pair. For proteins we abbreviate: D = Dalton; kD ~ kiloDalton.
Temperatures are always given in degrees Celsius.
B. Host Cell Cultures and Vectors 25 Useful recombinant human factor VIII may be produced, according to the present invention, in a variety of recombinant host cells. A
particularly preferred system is described herein.
In general, prokaryotes are preferred for cloning of DNA
sequences in constructing the vectors useful in the invention. For 30 example, E. coli K12 strain 294 (ATCC No. 31446) is particularly useful. Other microbial strains which may be used include E. coli strains such as E. coli B, and E. coli X1776 (ATTC No. 31537), and E. coli c600 and c600hf1, E. coli W3110 (F-, ~', prototrophic, ATTC Plo. 27325), bacilli such as Bacillus subtilus, and other 35 enterobacteriaceae such as Salmonella ty~himurium or Serratia -m-marcesans, and various pseudomonas species. These examples are, of course, intended to be illustrative rather than limiting.
In general, plasmid vectors containing replicon and control sequences which are derived from species compatible with the host cell are used in connection with these hosts. The vector ordinarily carries a replication site, as well as marking sequences which are capable of providing phenotypic selection in transformed cells. For exam!~le, E. coli is typically transformed using pBR322, a plasmid derived from an E. coli species (32). pBR322 contains genes for ampicillin and tetracycline resistance and thus provides easy means for identifying and selecting transformed cells. The pBR322 plasmid, or other microbial plasmid, irnzst= also contain, or be modified to contain, promoters which can be used by the microbial organism for expression of its own proteins. Those promoters most commonly used in recombinant ONA construction include the s-lactamase (penicillinase) and lactose promoter systems (33 - 35) and a tryptophan (trp) promoter system (36, 37). While these ar_e the most commonly used, other microbial promoters have been discovered and utilized, and details concerning their nucleotide sequences have been published, enabling a skilled worker to ligate them functionally with plasmid vectors (38).
In addition to prokaryotes, eukaryotic microbes, such as yeast cultures, may also be used. Saccharomyces cerevisiae, or common baker's yeast, is the most commonly used among eukaryotic microorganisms, although a number of other strains are commonly available. For expression in Saccharomyces, the plasmid YRp7, for example, (39 - 41) is commonly used. This plasmid already contains the tr~l gene which provides a selection marker for a mutant strain of yeast lacking the ability to grow in tryptophan, for example, AZ~CC
No. 44076 or PEP4-1 (42). The presence of the trill lesion as a characteristic of the yeast host cell genome then provides an effective environment for detecting transformation by growth in the absence of tryptophan.
Suitable promoting sequences in yeast vectors include the promoters for 3-phosphoglycerate kinase (43) or other glycolytic w 1 341395 -LO-enzymes (44, 45), such as enolase, glyceraldehyde-3-phosphate dehydrogenase, hexokinase, pyruvate decarboxylase, phosphofructokinase, glucose-6-phosphate isomerase, 3-phosphoglycerate mutase, pyruvate kinase, triosephosphate isomerase, phosphoglucose isamerase, and glucokinase. In constructing suitable expression plasmids, the termination sequences associated with these genes are also ligated into the expression vector 3' of the sequence desired to be expressed to provide polyadenylation of the mRNA and termination. Other promoters, which have the additional advantage of transcription controlled by growth conditions,are the promoter regions for alcohol dehydrogenase 2, isocytochrome G, acid phosphatase, degradative enzymes associated with nitrogen metabolism, and the aforementioned glyceraldehyde-3-phosphate dehydrogenase, and enzymes responsible for maltose and galactose utilization. Any plasmid vector containing yeast-compatible promoter, origin of replication and termination sequences is suitable.
Use of cultures of cells derived from multicellular organisms as cell hosts is preferred, particularly for expression of underlying DNA to produce the functional human factor VIII hereof, and reference is particularly had to the preferred embodiment hereof. In principle, vertebrate cells are of particular interest, such as UERO
and HeLa cells, Chinese hamster ovary (CHO) cell lines, and W138, BHK, COS-7 and ~~DCK cell lines. Expression vectors for such cells ordinarily include (if necessary) (an) origins) of replication, a promoter located in front of the gene to be ~lxpressed, along with any necessary ribosome binding sites, R1JA splice sites, polyadenylation site, and transcriptional terminator sequences.
For use in mammalian cells, the control functions on the expression vectors may be provided by viral material. For example, commonly used promoters are derived from polyoma, Simian Virus 40 (SY40) and most particularly Adenavirus ~'. hhe early and late promoters of SY40 virus are useful as is the major late promoter of adenovirus as described above. Further, it is also possible, and often desirable, to utilize promoter or contr°ol sequences normally _2I- ~ 3 41 3 9 5 associated with the desired gene sequence, provided such control sequences are compatible with the host cell systems.
An origin of replication may be provided either by construction of the vector to include an exogenous origin, such as may be derived from adenovirus or other viral (e. g. Polyoma, SV40, VSY, BPY, etc.) source, or may be provided by the host cell chromosomal replication mechanism, if the vector is integrated into the host cell chromosome.
In selecting a preferred host cell for transfectian by the vectors of the invention which comprise DNA sequences encoding both factor VIII and DHFR protein, it is appropriate to select the host according to the type of DHFR protein employed. If wild type DHFR
protein is employed, it is preferable to select a host cell which is deficient in DHFR, thus permitting the use of the DtiFR coding sequence as a marker for successful transfection in selective medium which lacks hypoxanthine, glycine, and thymidine.
On the other hand, if DHFR protein with low binding affinity for MTX is used as the controlling sequence, it is not necessary to use _ DHFR resistant cells. Because the mutant DHFR is resistant to methotrexate, h9TX containing media can be used as a means of selection provided that the host cells themselves are methotrexate sensitive. Most eukaryotic cells which are capable of absorbing MTX appear to be methotrexate sensitive.
Alternatively, a wild type DHFR gene may be employed as an amplification marker in a host: cell which is not deficient in DHFR
provided that a second drug selectable marker is employed, such as neomycin resistance.
Examples which are set forth hereinbelow describe use of BHK
cells as host cells and expressian vectors which include the adenovirus major late promoter.
C. General Methods If cells without formidable cell wall barriers are used as host cells, transfection is carried out by the calcium phosphate precipitation method (4S). However, other methods for introducing ONA into cells such as by nuclear injection or by protoplast fusion may also be used.
If prokaryotic cells or cells which contain substantial cell wall constructions are used, the preferred method of transfection is calcium treatment using calcium chloride (47).
Construction of suitable vectors containing the desired coding and control sequencese~loys standard ligation techniques. Isolated plasmids ar DNA fragments are cleaved, tailored, and religated in the form desired to form the plasmids required.
Cleavage is performed by treating with restriction enzyme (or enzymes) in suitable buffer. In general, about 1 up plasmid or DNA
fragments are used with ahoutl unit of enzyme in about 20 u1 of buffer solution for 1 haur. (Appropriate buffers and substrate amounts for particular restriction enzymes are specified by the manufacturer. Likewise, standard conditions for use of T4 lipase, T4 polynucleotide kinase and bacterial alkaline phosphatase are _ provided by the manufacturer.) After incubations, protein is removed by extraction with phenol and chloroform, and the nucleic acid is recovered from the aqueous fraction by precipitation with ethanol. Standard laboratory procedures are available (48).
Sticky ended (overhanging) restriction enzyme fragments are rendered blunt ended, far example, by either:
Fill in repair: 2-15 up of DNA were incubated in 50 mM NaCI, 10 mM Tris (pH 7.5), 1U mM DtgCl2, 1 mM dithiothreitol with 250 uM
each four deoxynucleoside triphosphates and 8 units DNA polymerase Klenow fragment at 24~C for 30 minutes. The reaction was terminated by phenol and chloroform extraction and ethanol precipitation, or S1 digestion: 2-15 up of DNA were incubated in 25 m~9 NaOAc (pH
4.5), 1 mM ZnCl2, 300 mM NaCI with 600 units Sl nuclease at 37' for 30 minutes, followed by phenol, chloroform and ethanol precipitation.
Synthetic DNA fragments were prepared by known phosphotriester (47a) or phosphoramidite (47b) procedures. DNA is subject to electrophoresis in agarose or polyacrylamide slab gels by standard procedures (48) and fragments were purified from gels by _~3_ 1 3 41 3 9 5 electroelution (48a). DNA "Southern" blot hybridization followed the (49a) procedure.
RtJA "Northern" blot hybridizations followed electrophoresis in agarose slab gels containing 5 percent formaldehyde. (48, 49b) Radiolabeled hybridization probes are prepared by random calf thymus DNA primed synthesis (49c) employing high specific activity 32P-labeled nucleotide triphosphates (32P: Amersham; Klenow DNA
polymerase: BRL, NEB or Boehringer-Mannheim). Short oligonucleotide probes may be end-labelled ~~rith T4 polynucleotide kinase. "Standard salt" Southern hybridization conditions ranged from: Hybridization in 5x SSC (lx SSC ~ 0.15 M NaCl 0.0I5 M Na3 citrate), 50 mM Na Phosphate pH 7, 10 percent dextran sulfate, 5x Denhardt's solution (lx Denhardt's = 0.02 percent ficoll, 0.02 percent polyvinylpyrrolidone, 0.02 percent bovine serum albumin), 20-100 ug/ml denatured salmon sperm DNA, 0-50 percent formamide at temperatures rangi ng frcxn 24° to 42° C, fol l owed by washes i n 0 .2-lx SSC plus 0.1 percent SDS at temperatures ranging from 24° -65°C.
Dried filters were exposed to Kodak XAR film using DuPont Lightning-Plus intensifying screens at -80~C. See, generally, (48).
For Northern blot screening of cell and tissue RNAs, hybridization was in 5x SSC, 5x Denhardt's solution, 10 percent dextran sulfate, 50 percent farmamide, 0.1 percent SDS, 0.1 percent sodium pyrophosphate, 0.1 mg/ml E. coli tRNA at 42~C overnight with 32P_labeled probe prepared from the 189 by StuI/HincI fragment of x120 containing exon A sequence. Wash conditions were 0.2x SSC, 0.1 percent SDS at 42°C.
Human DNA was prepared from peripheral blood lymphocytes (46,XY) or lymphoblast cells (49,XXXXY, N.I.G.M.S. Human Genetic Mutant Cell Repository, Camden, N.J., No. Gt~11202A) (48). E. coli plasmid DNA
was prepared as in (48) and bacteriophage a DNA (48). Tissue RNA
was prepared by either the guanidimium thiocyanate method (48, 49f) or the method of (49b). Polyadeny'lated RNA was isolated on oligo (dT) cellulose (49h).
DNA sequence analysis was performed by the method of (49i).

_24_ For the a/4X library, five 50 ug aliquots of the 49, XXXXY DNA
was digested in a 1 ml volume with S_au3AI concentrations of 3.12, 1.56, 0.782, 0.39, and 0.195 tl/ml for 1 hr at 37~C. Test digestion and gel analysis had shown that under these conditions at 0.782 ll/mi Sau3AI, the weight average size of the ONA was about 30 kb; thus these digests generate a number average distribution centered at 15 kb. DNA from 5 digests was pooled, phenol and chloroform extracted, ethanol precipitated and elect:rophoresed on a 6 g/1 low-gelling temperature horizontal agarose gel (48) (Seaplaque agaroseFh~C
Corporation), in two 5.6 x 0.6 x O.IS cm =slots. The 12-18 kb region of the gel was cut out and the DNA purified by melting the gel slice as described in (48).
Charon 30 arms were prepared by digesting 50 ug of the vector with BamHI and isolating the annealed 31.4 kb arm fragment from a 6 g/1 low-gelling temperature agarose gel as described above. For construction of the a/4X library, the optimal concentration of Charon 30 BamHI arms and 12-18 kb Sau3A partial 49,XXXXY DNA was determined as described (48). The ligated DhJA was packaged with an in vitro extract, "Packa~;ene;"* {Promega 8iotec, Inc., hladison, fI).
In a typical reaction about 1.:3 ug of~Gharon 30 Bamtll arms were ligated to 0.187 ug of 12-18 kt~ Sau3A insert Ut~A in a 10 u1 volume.
Packaging the plating of the DtdA gave about I.3x106 phage plaques. To generate the a4X library, 1.7x106 phage were plated at 17000 phage per 150 cm plate. These plates were grown overnight, scraped into 10 mh1 Tris HCI, pFi 7.5, 0.1 M NaCi, 10 mM MgCi2, 0.5 g/1 gelatin, and centrifuged briefly, to amplify the phage.
Generally, a suitable number (0.5-2x106) of these phage were plated out and screened {48). In some cases the ligated and in vitro packaged phage were screened directly without amplification.
For the isolation of x482, a clone containing a 22 kb BclI
fragment of the Factor VIII genome, <~nd t~tn~ ~~m ann fragments of the vector x1059 (49) were isolated by gel ele<:trop horesis. Separately, 100 ug of DNA from the 49,XXXXY cell line was digested with EiclI and the 20-24 kb fraction isolated by gel electrophoresis. About 0.8 ug of x1059 arms fragments and 5 percent of the isolated BclI DldA were *trade-mark .~

_25_ 1 3 41 3 9 5 ligated in a volume of 10 "1 (48) to generate 712,000 plaques. Four hundred thousand of those were screened in duplicate with 2.2 kb StuI/EcoRI probe of x114.
The cosmid/4X library was generated from the 49,XXXXY DNA used to generate the a/4X library, except that great care was used in the DNA isolation to avoid shearing or other breakage. The DNA was partially cleaved with five concentrations of Sau3AI and the pooled DNA sized on a 1U0 to 400 g/1 sucrose gradient (49). The fractions containing 35-45 kb DNA were pooled, dialyzed, and ethanol precipitated. Arm fragments of the cosmid vector pGeos4 were prepared following the principles described elsewhere (50). In brief, two separate, equal aliquots of pGcos4 were cut with SstI (an isoschizomer of SacI) or SaII and then treated with bacterial alkaline phosphatase. These aliquots were then phenol and chloroform extracted, pooled, ethanol precipitated and cut with BamHI. From this digest two arm fragments of 4394 and 4002 b were isolated from a low-gelling temperature agarose gel. These arm fragments were then ligated to the isolated, 40 kb Sau3AI partial digest DNA. In a typical reaction, 0.7 ug of pGcos4 arm fragments were ligated to 1 ug of 40 kb human 4X DNA in a volume of 10 u1 (48). This reaction was then packaged in vitro and used to infect E, coli HB101, a recA- strain (48). This reaction generated about 120,000 colonies when plated on tetracycline containing plates.
About 150,000 cosmids were screened on 20 150-mm plates in duplicate as described, with overnight amplificatian on chloramphenicol-containing plates (48).
Double-stranded cDNA was prepared as previously described (36, 67) employing either oligo(dT)12-18 or synthetic deoxyoligonucleotide 16-mers as primers for first-strand synthesis by reverse transcriptase. Following isolation by polyacrylamide gels, cDNA of the appropriate size (usually 600 by or greater) was either C-tailed with terminal transferase, annealed together with G-tailed PstI-digested pBR322 and transformed into E. coli strain DH1 (76), or ligated with a 100-fold molar excess of synthetic DNA
EcoRI adaptors, reisolated on a polyacrylamide gel, inserted by ligai;ion in EcoRI-digested aGTlO, packaged into phage particles and propagated on E. coli strain C600hf1 (68). As a modification of existing procedures an adaptor consisting of a complementary synthetic DP~A 18-mer and 22-mer (5'-CCTTGACCGTAAGACATG and 5'AATTCATGTCTTACGGTCAAGG) was phosphorylated at the blunt terminus but not at the EcoRI cohesive terminus to permit efficient ligation of the adaptor to double-stranded cDNA in the absence of extensive self-ligation at the EcoRI site. This effectively substituted for the more laborious procedure of ligating self-complementary EcoRI
linkers to EcoRI methylase-treated double-stranded cDNA, and subsequently removing excess linker oligomers from the cDNA termini by EcoRI digestion. To improve the efficiency of obtaining cDNA
clones >3500 by extending from the poly(A) to the nearest existing 3' factor VIII probe sequences made available by genomic cloning (i.e., exon A), second-strand cDNA synthesis was specifically primed by including in the reaction a synthetic DNA 16-mer corresponding to a sequence within exon B on the mRNA sense strand. _ D. Adenovirus Subcloning Adenovirus 2 DNA was purchased from Bethesda Research Laboratories (BRL). The viral DNA was cleaved with HindIII and electrophoresed through a 5 percent polya<:rylamide gel (TBE
buffer). The region of the gel containing the HindIII B fragment (49j) was excised and the ONA elect;roeluted from the gel. After phenol-chloroform extraction, the DNA was concentrated by ethanol precipitation and cloned into HindIII-cleaved pUCl3 (49k) to generate the plasmid pAdHindB. This HindIII subclone was digested with HindIII and SaII, and a fragment was isolated spanning adenoviral coordinates 17.1 - 25.9 (49j). This fragment was cloned into HindIII, SaII cleaved pUCl3 to generate the plasmid pUCHS.
From pAdHindB the SaII to XhoI fragment, coordinates 25.9 - 26.5, was isolated and cloned into pUCHS at the unique SaII site to create pUCI~SX. This plasmid reconstructs the adenoviral sequences from -z7- 1 3 4 ~ 3 9 5 position 17.1 within the first late leader intervening sequence to the XhoI site at position 26.5 within the third late leader exon.
The adenovirus major late promoter was cloned by excising the HindIII C, D, and E fragments (which comigrate) from the acrylamide gel, cloning them into pUCl3 at the HindIII site, and screening for recombinants containing the HindIII C fragment by restriction analysis. This subclone was digested with _SacI, which cleaved at position 15.4, 5' of the major late promoter {49j) as well as within the polylinker of pUCl3. The D1JA was recircularized to form pM~P2, containing the SacI to HindIII fragment (positions 15.4 - 17.1) cloned in the SacI and HindIII sites of pUCl3.
E. Construction of Neomycin Resistance Vector ~5 The neomycin resistance marker contained within E. coli transposon 5 was isolated from a Tn5 containing plasmid (491). The sequence of the neomycin resistance gene has been previously published (49m). The neo fragment was digested with Bc~III, which cleaves at a point 36 by 5' of the translational initiation codon of 20 the neomycin phosphotransferase gene, and treated with exonuclease Ba131. The phosphotransferase gene was excised with BamHI, which cleaves the DNA 342 by following the translational termination codon, and inserted into pBR322 between a filled-in HindIII site and the BamHI site. One clone, pPJeoQal6, had the translational 25 initiation codon situated 3 by 3' of the filled in HindIII site (TCATCGATAAGCTCGCATG...). This plasmid wa,s digested with CIaI and BamHI, whereupon the 1145 by fragment spanning the phosphotransferase gene was isolated and inserted into the mammalian expression vector pCVSVEHBS (see infra.). The resultant plasmid, 30 PSVENeoBal6, situates the neomycin phosphatransferase gene 3' of the SV40 early promoter and 5' of the polyadenylation site of the HBV
surface antigen gene (49n). When introduced into mammalian tissue culture cells, this plasmid is capable of expressing the phosphotransferase gene and conferring resistance to the 35 aminoglycoside 6418 (490).

-~8- 1 3 41 3 9 5 F. Transfection of Tissue Culture Celis The BHK-21 cells (ATCC) are vertebrate cells grown in tissue culture. These cells, as is known in the art, can be maintained as permanent cell lines prepared by successive serial transfers from isolated normal cells. These cell lines are maintained either on a solid support in liquid medium, or by growth in suspensions containing support nutrients.
The cells are transfected with 5 ug of desired vector (4 ug pAML3P.8c1 and 1 ug p~YEneoBal6) as prepared above using the method of (49p).
The method insures the interaction of a collection of plasmids with a particular host cell, thereby increasing the probability that if one plasmid is absorbed by a cell, additional plasmids would be absorbed as well (49q). Accordingiy, it is practicable to introduce both the primary and secondary coding sequences using separate vectors for each, as well as by using a single vector containing both sequences.
G. Growth of Transfected Cells and Expression of Peptides The BHK cells which were subjected to transfection as set forth above were first grown for two days in non-selective medium, then the cells were transferred into medium containing 6418 (400 ug/ml), thus selecting for cells which are abie to express the plasmid phosphotransferase. After 7-10 days in the presence of the 6418, colonies became visible to the naked eye. TrypsiniZation of the several hundred colonies and replacing allowed the rapid growth of a confluent 10 cm dish of 6418 resistant veils.
This cell population consists of cells representing a variety of initial integrants. In order to obtain cells which possessed the greatest number of copies of the F11III expression plasmid, the cells were next incubated with an inhibitor of the DHFR protein.
H. Treatment with Methotrexate The 6418 resistant cells are inhibited by methotrexate (MTX), a specific inhibitor of OHFR at concentrations greater than 50 nM.

-29- 1 3 ~+ 1 ~ 9 5 Consistent with previous studies on the effects of MTX on tissue culture cells, cells resistant to MTX by virtue of expression of the multiple copies of the DHFR gene contained within the FVIII
expression vector are selected for, and a concomitant increase in expression of the FYIII encoding sequences can be observed. By stepwise increasing the amount of MTX, amplification of the plasmid pAML3P.8c1 is affected, thus increasing the copy number. The upper limit of the amplification is dependent upon many factors, however cells resistant to millimolar concf5ntrations of MTX possessing hundreds or thousands of copies of the DHFR expression (and thus the FVIII expression) plasmid may be sE~lected in this manner.
For Factor VIII expression, 6418-resistant BHK cells which arose after transfection with pAML3P.8c1 and pSVENeo8a16 were incubated with media containing 100 nh1 and 250 nM hITX as described (49r).
After 7-10 days, cells resistant to 250 nP1 MTX were assayed for Factor VIII expression by activity, radioimmunoassay and mRNA
Northern analysis.
I. Factor VIII antibodies A variety of polyclonal and monoclonal antibodies to Factor VIII
were used throughout this work. CC is a polyclonal antibody derived from the plasma of a severely affected hemophiliac (49s). C8 is a neutralizing monoclonal antibody which binds to the 210 kD portion of Factor VIII (49t). C10 is a monoclonal antibody with properties similar to C8 and was isolated essentially as described by (49t). A
commercial neutralizing monoclonal antibady which binds the 80 kD
portion of Factor VIII was obtained from Synbiotic Corp., San Diego, CA., Product No. 10004. C7F7 is a neutralizing monoclonal antibody that binds to the 80 kD portion of Factor VIII. C7F7 was induced and purified as follows: Six-week-old female BALB/c mice were multiply inoculated with approximately 10 ug of purified Factor VIII
and splenocytes fused with X63-Ag8.653 mouse myeloma cells (49u) three days after the final inoculation. The hybridization procedure and isolation of hybrid cells by cloning methods followed previously described protocols (49r). Specific antibody producing v 1 341 39 5 _.
clones were detected by solid phase RIA procedures (49w). Positive clones were subsequently assayed far coagulation prolongation capacity by APTT assay described above. Monoclonal C7F7 was expanded by growth in syngeneic animals; antibody was purified from ascites fluids by protein A-Sepharose CL-4B*chromatography (49x).
J. Radioimmune Assays for Factor YIII
Two radioimmune assays (RIA) were developed to assay Factor VIII
produced from BHK and other cell lines. E3oth are two stage assays in which the CC antibody bound to a solid support is used to bind Factor VIII (49t). This immune complex is then detected with 1125 labeled C10 antibody (210 kD specific) or I125 labeled C7F7 antibody (80 kD specific).
Briefly, the two-stage RIAs are performed as follows: the 96 wells of a micratiter dish are coated overnight with 100 u1 of 50 mM
NaHC03 buffer, pH 9.6 containing 2.5 mg/1 of CC antibody which has been purified by protein A-sepharose chromatography (49x). _The wells are washed three times with 200 u1 of PBS containing 0.05 percent Tween 20 and blocked with 200 u1 of PBS containing O.I
percent gelatin and 0.01 percent met:.hiolate for 1 to 2 hours. The wells are washed as before and 100 ~1 of sample added and incubated overnight. The wells are washed and 100 4~1 of I125 labeled (82) C10 or C7F7 antibody (1000 cpm/ul) added and incubated 6 to 8 hours. The wells are washed again and counted. The standard curve is derived from samples of normal plasma diluted 1:10 to 1:320.
K. Factor VIII P1onoclonal Antibody Column A human factor VIII monoclonal antibody column was prepared by i ncubati on of 1.0 mg of C8 anti body ( i n ~) . 1h1 tJaHC03, pHB. 5) wi th 1.0 ml of Affi-Gel 10*(Bio-Rad Laboratories, Richmond, CA) for four hours at 4~ C. Greater than 95 percent of the antibody was coupled to the gel, as determined by the Bio-Rad Protein Assay (Bio-Rad Laboratories). The gel was washed with 50 volumes of water and 10 volumes of 0.05M imidazole, pH6.9, containing 0.15 h1 NaCI.
*trade-mark L . ~~oma ~ayr o_f M~d;_a ~~Pc>r7-~1a1 . ~l ~mn Media was applied to the monoclonal antibody column (lml.of resin) and washed with 0.05 M imidazole buffer, pH6.4, containing 0.15 M NaCI until material absorbing at 280 nm was washed off. The column was eluted with 0.05 M imidazole, pH6.4, containing 1.0 M KI
and 20 percent ethylene glycol. Samples were diluted for assay and dialized for subsequent analysis.
t~1. Preparation of Factor VIII Fusion Proteins and Fusion Protein Antisera _E. c_oli containing the plasmids constructed for fusion protein expression were grown in M-9 media at 3?°C. Fusion protein expression was induced by the addition of indole acrylic acid at a final concentration of 50 ug/mL for time periods of 2.5 to 4 hours.
The cells were harvested by centrifugation and frozen until use.
The cell pellets for fusion 3 were suspended in 100 ml of 20mM
sodium phosphate, pH 7.2, containing 10 ug/mL lysozyme and 1 ug/mL
each of RNase and DNase. The suspension was stirred for 30 minutes at room temperature to thoroughly disperse the cell pellet. The suspension was then sonicated for four minutes (pulsed at 60 percent power). The solution was centrifuged at 8000 rpm in a Sorvall RC-2B
centrifuge in a GSA rotor. The pellet was resuspended in 100 mL of 0.02 M sodium phosphate, pH ~.2. The suspension was layered over 300 mL of 60 percent glycerol. The sample was centrifuged at 4000 rpm for 20 minutes in an RC-3B centrifuc3e. Two layers resulted in the glycerol. Both pellet and the bottam glycerol layer showed a single protein band of the expected a~olecular weight of 25,000 daltons when analyzed on SOS polyacrylamide gels. The pellet was dissolved in 0.02 M sodium phosphate buffer containing 0.1 percent SDS. The resuspended pellet and the lower glycerol layer were dialyzed against 0.02 M ammonium bicarbonate, pH 8.0, to remove glycerol. The solution was lyophilized and redissolved in 0.01 t~1 sodium phosphate buffer containing 0.1 percent SOS, and frozen until use.

The cell pellets for fusion proteins 1 and 4 were suspended in 0.05 M Tris, pH 7.2, containing U.3M sodium chloride and 5 mM EDTA.
Lysozyme was added to a concentration of 10 ug/mL. Samples were incubated for 5 minutes at room temperature. NP-40 was added to 0.2 percent and the suspension incubated in ice for 30 minutes. Sodium chloride was added to yield a final concentration of 3M and DNase added (1 ug/mL). The suspension was incubated 5 minutes at room temperature. The sample was centrifuged and the supernatant discarded. The pellet was resuspended in a small volume of water 1U and recentrifuged. The cell pellets were dissolved in solutions containing 0.1 percent to 1 percent SOS and purified by either preparative SDS polyacrylamide gel electrophoresis followed by electroelution of the fusion protein band, or by HPLC on a TSK 3000 column equilibrated with O.lh1 sodium phosphate containing 0.1 percent SDS.
rabbit antisera were produced by injecting New Zealand white rabbits with a sample of fusion protein suspended in Freund's complete adjuvant (first injection) followed by boosts at two week intervals using the sample suspended in Freund's incomplete 2U adjuvant. After six weeks, sera were obtainE~d and analyzed by Western Blot analysis for reactivity with human plasma derived factor VIII proteins.
N. Assays for Detection of Expression of Factor VIII Activity Correction of Hemophilia,A plasma. - Theory - Factor VIII
activity is defined as that activity which will correct the coagulation defect of factor VIII deficient plasma. One unit of factor VIII activity has been defined as that activity present in one milliliter of normal human plasma. The assay is based on 3U observing the time required for formation of a visible fibrin clot in plasma derived from a patient diagnosed as suffering from hemophilia A (classic hemophilia). In this assay, the shorter the time required for clot formation, the greater the factor VIII
activity in the sample being tested. This type of assay is referred to as activated partial thromboplastin time (APTT). (:ommercial _33_ 1 3 41 3 9 5 reagents are available for such determinations (for example, General Diagnostics Platelin Plus Activator; product number 35503).
Procedure - All coagulation assays were conducted in IO x 75 mm borosilicate glass test tubes. Siliconization was performed using SurfaSil*(product of Pierce Chemical Company, Rackford, IL) which had been diluted 1 to IO with petroleum ether. The test tubes were filled with this solution, incubated 15 seconds, and the solution removed. The tubes were washed three times with tap water and three times with distilled water.
Platelin Plus Activator (General Diagnostics, Morris Plains, NJ) was dissolved in 2.5 ml of distilled water according to the directions on the packet. To prepare the sample for coagulation assays, the Platelin plus Activator solution was incubated at 37°C
for 10 minutes and stored tin ice until use. To a siliconized test tube was added 50 microliters of Platelin plus Activator and 50 microliters of factor VIII deficient plasma (George King Biomedical Inc, Overland Park, KA). This solution was incubated at 37°C for a total of nine minutes. Just prior to the end of the nine minute incubation of the above solution, the sample to be tested was diluted into 0.05 M Tris-HC1, pH 7.3, captaining 0.02 percent bovine serum albumin. To the plasma/activator suspension was added 50 microliters of the diluted sample, and, at exactly nine minutes into the incubation of the suspension, the coagulation cascade was initiated by the addition of 50 microliters of calcium chlaride (0.033 M). The reaction mixture was quickly mixed and, with gentle agitation of the test tube, the time required for the formation of a visible fibrin clot to form was monitored. A standard curve of factor VIII activity can be obtained by diluting normal plasma (George King Biomedical, Inc., Overland Park, KA) 1:10, 1:20, 1:50, 1:100, and 1:200. The clotting time is platted versus plasma dilution on semilog graph paper. This can then be used to convert a clotting time into units of factor VIII activity.
0. Chromogenic Peptide Determination Theory - Factor VIII functions in the activation of factor X to factor xa in the presence of factor zx~, phospholipid, and calcium *trade-mark 1 34~ 39 5 ions. A highly specific assay has been designed wherein factor Ixa, factor X, phospholipid, and calcium ions are supplied. The generation of factor xa in this assay is therefore dependent upon the addition of a source of factor VIII activity. The more factor VIII added to the assay, the more factor xa is generated. After allowing the generation of factor xa, a chromogenic peptide substrate is added to the reaction mixture. This peptide is specifically cleaved by fiactor xar is nr~t effected by factor X, and is only slowly cleaved by other proteases. Cleavage of the peptide substrate releases a para-vitro-anilide grouE:a which has absorbance at 405 nm, while the uncleaved peptide substrate has little or no absorbance at this wavelength. The generation of absorbance due to cleavage of the chromogenic substrate is dependent upon the amount of factor xain the test mixture after the incubation period, the 1~i amount of which is in turn dependent upon the amount of functional factor VIII in the test sample added to the reaction mixture. This assay is extremely specific for factor VIII activity and should be less subject to potential false positives when compared to factor VIII deficient plasma assay.
Procedure - Coatest factor VIIT was purchased from Helena Laboratories, Beaumont, TX (Cat. No. 5253). The basic procedure used was essentially that provided by th a manufacturer for the "End Point t~iethod" for samples containing less than 5 percent factor VIII, Where indicated, the times of incubation were prolonged in 2r order to make the assay more sensitive. For certain assays the volumes Of reagents recommended b~ the manufacturer ware altered.
This change in the protocol does not interfere with the overall results of the assay.
The chromogenic substrate (S-2222 ~ I-2581) for factor xa was dissolved in 10 milliliters of water, resulting in a substrate concentration of 2.7 millimoles per liter. This substrate solution was aliquoted and stored frozen at -20QC. The Flxa + FX reagent contained the factor zxa and factor X and was dissolved in 10 milliliters of water. The solution was aliquoted and stored frozen at -70~C until use. Also supplied with the kit were the following _3,_ 1 3 41 3 9 5 solutions: 0.025 molar calcium chloride; phospholipid (porcine brain); and Buffer Stock Solution (diluted one part of Stock Solution to nine parts of water for the assay, resulting in a final concentration of 0.05 hi Tris-HCI, pH 7.3, containing 0.02 percent bovine albumin). These solutions were stored at 4°C until use.
The phospholipid + FIXa + FX reagent is prepared by mixing one vol urne of phosphol l pi d wi th fl ve vol umes of' i~'IXa+ FX reagent.
The following procedure was employed:
~0 Reagent Sample Tube Reagent Blank Phospholipid + FI~;a+ FX 200u1 200u1 Test sample 100 --Buffer working solution --- 100 Mix well and incubate at 37°C for 4 minutes Calcium chloride 100 100 Mix well and incubate at 37~C exactly 10 minutes S-2222 + I-2581 200 200 P~lix well and incubate at 37°G exactly 10 minutes Acetic acid (50 percent) I00 100 Mix well The absorbance of the sample at 405 nm was determined against the reagent blank in a spectrophotometer within 30 minutes.
The absorbance at 405 was related to factor VIII units by calibrating the assay using a standard normal human plasma (George King Biomedical, Overland Park, KS).
Example of Preferred Embodiment 1. General Strategy for Obtaining the Factor VIII Gene The most common process of obtaining a recombinant DNA gene product is to screen libraries of cDNA clones obtained from mRtdA of 04571.

. 1 341 39 5 the appropriate tissue or cell type. Several factors contributed to use also of an alternative method of screening genomic DNA for the factor VIII gene. First, the site of synthesis of factor VIII was unknown. Although the liver is frequently considered the most likely source of synthesis, the evidence is ambiguous. Synthesis in liver and possibly spleen have been suggested by organ perfusion and transplantation studies (56). However, factor VIII activity is often increased in patients with severe liver failure (56a). Recent conflicting studies employing monoclonal antibody binding to cells 1U detect highest levels of the protein in either l9ver sinusoidal endothelial (51), hepatocyte (52) or lymph node cells (followed in amount by lung, liver and spleen; (53)). In contrast, the factor VIII related antigen Cvon Wi~lebrand Factor) is almost certainly synthesized by endothelial cells (54). Not only is the tissue 1G; source uncertain, the quantity of factor YTII in plasma is extremely low. The circulating concentration of about 100-200 ngJml (55) is about 1/2,000,000 the molar concentration of serum albumin, for example. Thus, it was not clear that cUNA libraries made from RNA
of a given tissue would yield factor VIII clones.
2U Based on these considerations, it was decided to first screen recombinant libraries of the human genome in bacteriophage lambda (henceforth referred to as genomic libraries). Although genomic libraries should contain the factor VIII gene, the likely presence of introns might present obstacles to the ultimate expression of the 25 recombinant protein. The general strategy was to:
1. Identify a genomic clone corresponding to a sequenced portion of the human factor VIII protein.
2. Conduct a "genomic walk" to obtain overlapping genomic clones that would include the entire mRNA coding region.
30 3~ Use fragments of the genomic clones to identify by hybridization to RNA blots tissue or cell sources of factor YIII mRtJA and then proceed to obtain cDNA clones from such cells.
4. In parallel with no. 3, to express portions of genomic 3~ clones in SY40 recombinant "exon expression" plasmids. RNA

- 1 34~ 39 5 _37_ transcribed from these plasmids after transfection of tissue culture (cos) cells should be spliced in vivo and would be an alternative source of cD~dA clones suitable for recombinant factor VIII protein expression.
The actual progress of this endeavor involved simultaneous interplay of informatian derived from cDNA clones, genomic clones of several types, and SV40 recombinant "exon expression" clones, which, of necessity, are described separately below.
IQ 2. Genomic Library Screening Procedures The factor VIII gene is known to reside on the human X
chromosome (56). To increase the proportion of positive clones, genomic libraries were constructed from DNA obtained from an individual containing 4X chromosomes. (The lymphoblast cell line is karyotyped 49,XXXXY; libraries constructed from this DNA are referred to herein as "'4X libraries"). 49,XXXXY DNA was partially digested with Sau3AI and appropriate size fractions were ligated into a phage or cosmid vectors. Oetaiis of the construction of these a/4X and cosmid/4X libraries are given below. 'the expected 20 frequency of the factor VIII gene in the a/4X library is about one in 110,000 clones and in the cosmid library about one in 40,000.
These libraries were screened for the factor VIII gene with synthetic oligonucleotide probes based on portions of the factor VIII protein sequence. These oligonucleotide probes fall into two 25 types, a single sequence of 30 to 100 nucleatides based on codon choice usage analysis (long probes) and a pool of probes 14-20 nucleotides long specifying all possible degeneracy combinations for each codon choice (short probes).
The main advantage of long probes is that they ca.n be 30 synthesized based on any 10-30 amino acid sequence of the protein.
No special regions of low codon redundancy need be faund. Another advantage is that since an exact match with the gene sequence is not necessary (only stretches of complementarity of 10-14 nucleotides are required), interruption of complementarity due to presence of an 35 intron, or caused by gene palyrnorphism or protein sequencing error, _~~_ 1 3 41 3 9 5 would not necessarily prevent usable hybridization. The disadvantage of long probes is that only one codon is selected for each amino acid. We have based our choice of codons on a table of mammalian codon frequency (57), and when this gave no clear preference, on the codon usage of the Factor IX gene (58). Since the expected sequence match of the long probes is unknown, the hybridization stringency must be determined empirically for each probe. This was performed by hybridization to genomic DNA blots and washes at various stringencies.
The advantage of short probes is that every codon possible is synthesized as a pool of oligonucleotides. Thus if the amino acid sequence is correct, a short probe should always hybridize to the gene of interest. The main 'limitation is the complexity of the pool of sequences that can be synthesized. Operationally a pool of 32 16 different sequences might be considered as a maximum pool size given the signal to noise limitations of hybridization to genomic libraries. This means that only protein sequences in regions of low codon redundancy can be used. A typical prabe would be a pool of 16 17-mers specifying all possible sequences over a 6 amino acid fragment of protein sequence.
As with long probes, the hybridization stringency used for short probes had been determined empirically. This is because under ordinarily used hybridization conditions (6xSSC), the stability of the hybrids depends on the two factors--the length and the G-C
content; stringent conditions for the low G-C content probes are not at all stringent for the high G-C content ones. A typical pool of 16 17-mers might have a range of 4I to 65 percent Ca-C and these probes will melt in 6xSSC over a 10~C temperature range (from 48' -58~C). Since the correct sequence within the pool of 16 is not known in advance, one uses a hybridization stringency just below 48'C to allow hybridization c>f the lowest G-C content sequence.
However, when screening a large number of clones, this will give many false positives of shorter length and higher G-C content.
Since the change in melting temperature is 1 to 2~C per base pair match, probe sequences as short as 12 or 13 of the 17 will also bind -'. 1341395 if they have a high G-C content. At random in the human genome a pooled probe of 16 17-mers will hybridize with 1200 times as many 13 base sequences as 17 base sequences.
A hybridization technique was developed for short probes which equalizes the stability of G-C and A-T base pairs and greatly enhances the utility of usincj short probes to screen libraries of high DNA sequence complexity.
In Figure 2A is p'~tted the melting temperature of 4 short probes under ordinary (6xSSC~ and 3.0 M TMAC1 wash conditions. In 3.0 M TMAC1 the probes melt as a nearly linear function of length, while in 6xSSC, the melting is greatly influenced by 'the G-C
content. The high melting temperature in 6xSSC of the 13-mer that is 65 percent G-C clearly den3onstrates this conclusion. Figure 2B
shows the melting temperature in 3.0 M TMAC1 as a function of length for 11 to thousands of bases. This figure allows the rapid selection of hybridization conditions for a probe with an exact match of any length desired.
The TMAC1 hybridization procedure has great utility whenever an exact sequence match of some known length is desired. Examples of this technique include: 1. Screening of a human genomic library with a pool of 16 17-mers. lde have used a 3.0 M Tf4AC1 wash at 50'C, which allows hybridization of only 17, 16, and a few 15 base sequences. The large number of high G-C content probes of lower homology are thus excluded. 2. If a short probe screen yields too many positives to sequence easily, the mostly likely candidates can be found by a TP1AC1 melting procedure. Replicas of the positives are hybridized and washed at 2~C intervals (for 17-mer~s (which melt at 54~Cj 46, 48, 50, 52, 54, and 564C would be usedl. The positives that melt at the highest temperature will match the probe most closely. With a standard of known sequence the homology can be predicted +1 base or better for a 17-mer. 3. Similarly, if a long probe screen yields too many positives, pooled short probes based on the same protein sequence can be synthesized. Since one member of this pool would contain a perfect match, TMAC1 melting experiments could refine the choice of best candidate positives. 4. In site -40- 1 3 4 '! 3 9 5 directed mutagenesis, an oligonucleatide typically 20 long with 1 or more changes in the center is synthesized. The TMAC1 wash procedure can easily distinguish the parental and mutant derivatives even for a 1 base mismatch in the middle of a 20-mer. This is because the desired mutation matches the probe exactly. The wash conditions can simply be determined from figure 2t3. 5. Selection of one particular gene out of a family of closely related genes. A melting experiment similar to that described above teas been used to select one particular gene out of a collection of 100 very similar sequences.
3. First Isolation of the Factor VIII Genomic Clone Factor YITI enriched preparations were prepared from human cryoprecipitate by polyelectrolyte chromatography and immunoadsorption as previously described (79). This material was dialyzed into 0.1 percent sodium dodecyl sulfate (SDS) and 1 percent ammonium bicarbonate, lyophilized, and stored at -20'C until use.
Due to contamination of the factor VIII preparations by other plasma proteins, further fractionation was required in order to purify the factor VIII as wall as separate the various polypeptide chains believed to arise from the factor yIII. This was accomplished by chromatography of the protein on Toya Soda TSK 4000 SW columns using high pressure liquid chromatography in the presence of SDS. Such chromatography separates the proteins by molecular size.
The lyophilized protein was reconstituted in distilled water and made 1 percent SDS and 0.1 N sodium phosphate, pH 7.5. The TSK
column (0.75 x 50 cm; Alltech, Deerfield, TL) was equilibrated at room temperature with O.I percent SDS in O.I M sodium phosphate, pH
7Ø Samples of approximately 0.15 to 0.25 mL were injected and the column was developed isocratically at a fl«w rate of 0.5 mL per minute.
The absorbance was monitored at 280 nm and fractions of 0.2 mL were collected. A representative elution profile is shown in Figure 15.
Aliquots were analyzed by sodium dodecyl sulfate gel electrophoresis on gradient gels of 5 percent to 10 percent polyacrylamide and _41_ analyzed by silver staining (80). The material which eluted after 25 minutes corresponded to a doublet of proteins at 80,000 and 78,000 D. The fractions containing these proteins were pooled as indicated by bar in Figure 15, from three separate preparative TSK
runs, and stored at -20 degrees until use.
The purified 80,00() dalton protein from the TSK fractionation (0.8 nmoles) was dialyzed overnight against 8 hl urea, 0.36 M
Tris-HC1, pH 8.6, and 3.3 mM ethylenediamine-tetraacetic acid under a nitrogen atmosphere. Disulfide bonds were reduced by the inclusion of lOmM dithiothreitol in the above dialysis buffer. The final volume was 1,5 ml. The cysteines were alkylated with 15 microliters of 5 M iodoacetic acid (dissolved in 1M NaOH). The reaction was allowed to proceed for 35 minutes at room temperature in the dark, and the alkylation reaction was quenched by the addition of dithiothreitol to a final concentration of 100 mM. The protein solution was dialyzed against 8-M urea in 0.1 M ammonium bicarbonate for four haurs. The dialysis solution was changed to gradual 1y di l ute the urea concentrati on ( 8 h1, 4 P1, 2 h1, 1 M, and 20 finally 0.5 t~1 urea) over a period of 24 hours. Tryptic digestion was performed on the reduced, alkylated 80,000 dalton protein by the addition of TPCK-treated trypsin (Sigma Chem. Go.) at a weight ratio of 1 part trypsin to 3U parts factor VIII protein. The digestion was allowed to continue for L2 hours at 37~C. The reaction mixture 25 was frozen until use. HPLC separation of the tryptic peptides was performed on a high resolution Synchropak RP-P C-18 column (0.46 x 25 cm, 10 microns) at room temperature with a Spectra-Physics 8000 chromatograph. Samples of approximately 0.8 mL were injected and the colurm was developed with a gradient. of acetonit:rile (1 percent to 30 70 percent in 200 minutes) in 0.1 percent trifluoroacetic acid. The absorbance was monitored at 210 nm and 280 nm (Figure 16). Each peak was collected and stored at 4QC until subjected to sequence analysis in a Beckman spinning cup sequencer with on-line PTH amino acid identification. The arrow in Figure 16, eluting at 35 approximately 23 percent acetonitrile, indicates the peak containing the peptide with the sequence AWAYFSDYDLEK. This sequence was used to generate the oligonucleotide probe 8.3 for human genomic library screening.
Long and short probes were synthesized based on the considerations just discussed. The second long probe used was based on the sequence of a 12 amino acid factor VIII tryptic fragment, AWAYFSDVDLEK. The DNA sequence chosen to synthesize for this probe was 5'-CTTTTCCAGGTCAACGTCGGAGAAATAAGCCCAAGC. This probe (called 8.3) was first tested in genomic blot hybridizations. Figure 3A shows genomic Southern blots of normal male (1X) and 49,XXXXY (4X) DNA
hybridized with labeled 8.3 probe and washed at various stringencies. Even at the highest stringency (lxSSC, 46°C) a single band of 3.8 kb (EcoRI) and 9.4 kb (BamHI) was observed. The intensity of this band had a ratio of about 1:4 in the 1X and 4X
lanes as would be expected for the X-linked factor VIII gene.
Control experiments had demonstrated that a known X-linked gene probe (Factor IX) gave the expected 1:4 hybridization ratio, while an autosomal gene (albumin) gave a 1:l ratio.
Based on these genomic blot results, the 8.3 probe was used to 2p screen the a/4X library. 500,000 phage were grown on fifty 150 mm plates and duplicate nitrocellulose filters were hybridized with 32p_labeled 8.3 probe at a wash stringency of lxSSC, 37°C (Figure 3). Upon retesting, 15 strongly hybridizing and 15 more weakly hybridizing clones were obtained. DNA was prepared from these isolated plaques, cleaved with restriction endonucleases, and blot hybridized with probe 8.3. Many of the strongly hybridizing clones yielded a hybridizing EcoRI fragment of 3.8 kb, the same size detected in the genomic blot. In addition, all strongly hybridizing clones displayed an identical 262 base pair Sau3AI fragment upon 30 hybridization with the 8.3 probe. Sau3AI fragments were cloned into the single-stranded phage vector M13mp8 (86), screened by hybridization, and sequenced by the dideoxy procedure. The DNA
sequence of the 262 by fragment showed considerable homology with the 8.3 probe. The homology included regions of continuous matches 3G of 14 and 10 by with an overall homology of 83 percent. The first ten residues of the peptide fragment agreed with that deduced from the DNA sequence of the recombinant clones and they were preceded by a lysine codon as expected for the product of a tryptic digest. The final two predicted residues did not match the DNA sequence. However, the DNA at this juncture contained a goad consensus RNA splice donor sequence (60, 61) followed shortly by stop codons in all three possible reading frames. This suggested the presence of an intron beginning at this position. (This suggestion was confirmed with cDNA
clones described below.) An open reading frame extended almost 400 b 5' of the region of homology. In this region several consensus splice acceptor sequences were identified. Inspection of the DNA-predicted protein sequence for this region revealed matches with protein sequence of several additional tryptic peptide fragments of factor VIII. This demonstrated that an exon of a genomic clone for human factor YIII had been obtained.
4. Extension of Genomic Clones: a Library Genome Walking Initially 8 independent factor VIII genomic clones were obtained from the a/4X library. These contained overlapping segments of the human genome spanning about 28 kb. From the estimated size of the factor YIII protein, it was assumed that the complete gene would encompass 100-200 kb, depending on the length of introns. Hence the collection of overlapping clones was expanded by "genome walking".
The first step in this process was the mapping of restriction 2G> endonuclease cleavage sites in the existing genomic clones (Figure 4).
DNA from the clones was digested with restriction enzymes singly or in combinations, and characterized by gel electrophoresis (followed by Southern blot hybridization in some cases). DtJA fragments generated by EcoRI and BamHI digestion were subclone<i into pUC plasmid vectors (59) for convenience. Restriction mapping, DNA sequence analysis, and blot hybridizations with the 8.3 probe determined the gene orientation.
Next, single copy fragments near the ends of the 28 kb region were identified as "walk" probes. Digests of cloned DNA were blot hybridized with total 32P-labeled human DNA. With this technique only fragments containing sequences repeated more than about 50 ,y 1 341 39 5 times in the genome will hybridize (87, 88). Non-hybridizing candidate walk probe fragments were retested for repeated sequences by hybridization to 50,000 phage from the a/4X library.
In the 5' direction, a triplet of 1 kb probe fragments was isolated from x120 DNA digested with NdeI and BamHI (see Figure 4).
One million a/4X bacteriophage were screened with this probe. A
resulting clone, x222, was shown to extend about 13 kb 5' of x120 (see Fig. 4). , In the 3' direction, a 2.5 kb StuI/EcoRI restriction fragment of x114 was identified as a single copy walk probe. Exhaustive screening of the a/4X, and subsequently other ;/human genomic libraries, failed to yield extending clones. Under-representation of genomic regions in a libraries has been observed before (62). It was decided to specifically enrich genomic DNA for the desired sequences and construct from it a limited bacteriophage library.
Southern blot hybridization of human genamic DNA with the 2.5 kb StuI/EcoRI probe showed a 22 kb hybridizing BcII restriction fragment. Restriction mapping showed that cloning and recovery of this fragment would result in a large 3' extension of genomic clones. Human 49,XXXXY DNA was digested with BcII, and a size fraction of about 22 kb was purified by gel electrophoresis. This DNA was ligated into the BamHI site of the bacteriophage vector x1059 and a library was prepared. (The previously used vector, Charon 30, could not accommodate such a large insert.) Six hybridizing clones were obtained from 400,000 phage screened from this enriched library. The desired clone, designated x482, extended 17 kb further 3' than our original set of overlapping genomic clones (Fig. 4), 5. Genome Walking Cosmid Clones A new genomic library was constructed with cosmid vectors.
Cosmids (63), a plasmid and bacteriophagt~ hybrid, can accommodate approximately 45 kb of insert, about a three-fold increase over the average insert size of the al4X ONA library. A newly constructed 3G; cosmid vector, pGcos4, has the following desirable attributes:

1. A derivative of the tetracycline resistance gene of pBR322 was used that did not contain a BamHI site. This allowed a BamHI-site to be put elsewhere in the plasmid and to be used as the cloning site. Tetracycline resistance is somewhat easier to work with than a the more commonly used ampicillin resistance due to the greater stability of the drug. 2, The 403 b NincII fragment of a containing the cos site was substituted for the 641 b Aval/PvuII
fragment of pBR322 so that the copy number of the plasmid would be increased and to remove pBR322 sequences which interfere with the transformation of eukaryotic cells (757, 3. A mutant dihydrofolate reductase gene with an SY40 origin of replication and promoter was included in the pGcos4 vector. In this way any fragments cloned in this vector could then be propagated in a wide range of eucaryotic cells. It was expected this might prove useful in expressing large 1a fragments of genomic DNA with their natural promoters. 4. For the cloning site, a synthetic 20-mer with the restriction sites EcoRI, PvuI, BamNI, PvuI, and EcoRI was cloned into the EcoRI site from pBR322. The unique BamHI site is used to clone 35-45 b Sau3A1 fragments of genomic DNA. The flanking EcoRI sites can be used for 2U subcloning the EcoRI fragments of the insert. The PvuI sites can be used to cut out the entire insert in most cases. Pvul sites are exceedingly rare in eucaryotic DNA and are expected to occur only once every 134,000 b based on dinucieotide frequencies of human D~JA.
25 Figure 5 gives the scheme for constructing the cosmid vector, pGcos4. 35-45 kb Sau3A1 fragments of 49,XXXXY DNA were cloned in this vector. About 150,000 recombinants were screened in duplicate with a 5' 2.4 kb EcoRIIBamHI fragment of x222 and a 3' 1 kb EcoRI/BamHI fragment of x482 which were single copy probes 3U identified near the ends of the existing genomic region. Four positive cosmid clones were isolated and mapped. Figure 4 includes cosmids p541, p542 and p543. From this screen, these cosmid clones extended the factor VIII genomic region to a total of 114 kbp.
Subsequent probing with cDNA clones identified numerous exons in the existing set of overlapping genomic clones, but indicated that the .. -46- ~ 3 41 3 9 5 genomic walk was not yet complete. Additional steps were taken in either direction.
A 3' walk probe was prepared from a 1.1 kb BamHI/EcoRI fragment of p542 (Fig. 4). This probe detected the overlapping cosmid clone '' p6i3 extending about 35 kb farther 3'. At a later time, the full Factor VIII message sequence was obtained by cDtdA cloning (see below). When a 1.9 kb _EcoRI cDNA fragment containing the 3'-terminal portion of the cDNA was hybridized to Southern blots of human genomic and cosmid cloned DNA, it identified a single 4.9 kb EcoRI band and 5.7, 3.2 and 0.2 kb E3amHI bands in both noncloned (genomic) and p613 DNA. This implied that the 3' end of the gene had now been reached, as we later confirmed by DNA sequence analysis.
A 5' walk probe was prepared from a 0.9 kb EcoRi/BamHI fragment of p543. It detected an overlapping cosmid clone p612, which slightly extended the overlapping region. The 5'-most genomic clones were finally obtained by screening cosmid/4X and a/4X
- libraries with cDNA derived probes. As shown in Figure 4, X599, x605 and p624 complete the set of recombinant clones spanning Factor VIII gene. (These clones overlap and contain all of the DNA of this region of the human genome with the exception of an 8.4 kb gap between p624 and x599 consisting solely of intron DNA.) Together, the gene spans 200 kb of the human X chromosome. This is by far the largest gene yet reported. Roughly 95 percent of the gene is comprised of introns which must be properly processed to produce 2Ei template mRNA far the synthesis of Factor VIII protein.
The isolation of the factor VIII gene region in x and cosmid recombinant clones is not sufficient to produce a useful product, the factor VIII protein. Several approaches were followed to identify and characterize the protein coding (exon) portions of the gene in order to ultimately construct a recombinant expression plasmid capable of directing the synthesis of active factor VIII
protein in transfected microorganisms or tissue culture cells. Two strategies failed to yield substantially useful results: further screening of genomic clones with new oligonucleotide probes based on protein sequencing, and the use of selected fragments. of genornic 1 3'~1 39 5 clones as probes to RNA blot hybridizations. However, coding regions for the factor VIII protein were isolated with the use of SV40 "exon expression" vectors, and, ultimately, by cDNA cloning.

6. SY40 exon expression vectors It is highly unlikely that a genomic region of several hundred kb could be completely characterized by DNA sequence analysis or directly used to sy~thesize useful amounts of factor VIII protein.
Roughly 95 percent of the hurrran factor VIII gene comprises introns 1U (intervening sequences) which must be removed artificially or by eukaryotic RNA splicing machinery before the protein could be expressed. A procedure was created to remove introns from incompletely characterized restriction fragments of genomic clones using what we call SV40 expression vectars. The general concept 1E~ entails inserting fragments of genomic DNA into plasmids containing an SV40 promoter and producing significant amounts of recombinant RNA which would be processed in the transfected monkey cos cells.
The resulting spliced RNA can be analyzed directly or provide material for cONA cloning. In theory at least, this technique could 20 be used to assemble an entire spliced version of the factor VIII
gene.
Our first exon expression constructions used existing SV40 cDNA
vectors that expressed the hepatitis surface antigen gene (73).
However, the genomic factor VIII fragments cloned into these vectors gave no observable factor VIII RNA when analyzed by blot hybridization. It was surmised that the difficulty might be that in the course of these constructions the exon regions of the cDNA
vectors had been joined to intron regions of the factor VIII gene.
To circumvent these difficulties, the exon expression vector pESYDA
3U was constructed as shown in figure 6. This vector contains the SV40 early promoter, the Adenovirus II major late first splice donor site, intron sequences into which the genomic factor VIII fragments could be cloned, followed by the Adenovirus II Elb splice acceptor site and the hepatitis 8 surface antigen 3' untranslat:ed and polyadenylation sequences (49j).

_4$_ ~ 3 4 1 3 9 5 Initially the 9.4 kb BamHI fragment and the 12.7 kb SstI
fragment of x114 were cloned in the intron region of pESVDA (see Fig. 6). Northern blot analysis of the RNA synthesized by these two constructions after transfection of cos cells is shown in figure 7.
With the 9.4 kb BamHI construction, a hybridizing RNA band of about 1.8 kb is found with probes for axon A, and hepatitis 3' untranslated sequence. To examine the RNA for any new factor VIII
axons, a 2.0 kbp StuI/BamHI fragment of x114, 3' of axon A, was hybridized in a parallel lane. This probe also showed an RNA band of 1.8 kb demonstrating the presence of additional new factor VIII
axons in this region. Each of these three probes also hybridized to an RNA band from a construction containing the 12.7 k:b SstI genomic fragment. This RNA band was about 2.1 kb. This observation suggested that an additional 200-300 by of axon sequences were contained in this construction 3' of the BamHI site bordering the 9.4 kb BamHI fragment.
Control experiments showed that this system is capable of correctly splicing known axon regions. A 3,2 kb genomic HindIII
fragment of murine dhfr spanning axons III and IY was cloned in pESVDA. An RIJA band of 1 kb was found with a murine dhfr probe.
This is the size expected if the axons are spliced correctly.
Constructions with the 9.4 kb BamHI factor VIII or 3.2 kb dhfr genomic fragments in the opposite orientation, gave no observable RNA bands with any of the probes (Fig. 7).
A cDNA copy of the RNA from the 12.7 kb SstI construction was cloned in pBR322 and screened. One nearly full length (1700 bp) cDNA clone (S36) was found. The sequence of the 950 by SstI
fragment containing all of the factor VIII insert and a portion of the pESVDA vector on either side is presented in Figure 8. The sequence begins and ends witha the Adenavirus splice donor and acceptor sequences as expected. In between there are 888 by of factor VIII sequence includir;g axon A. The 154 by preceding and the 568 by following axon A contain several factor VIII 80K tryptic fragments, confirming that these are newly identified axons.
Sequences of the genomic region corresponding to these axons showed that the 154 by 5' of axon A are contained in one axon, C, and that the region 3' of axon A is composed of 3 axons, D, E, and I of 229, 1$3 and 156 by respectively. Each of these axons is bounded by a reasonable splice donor and acceptor site t60, 61).
Subsequent comparison of the S36 axon expression cDNA with the factor VIII cell line cDtdA clones showed that all the spliced factor VIII sequence in S36 is from factor VIII axons. This included as expected axons C, A, D, E, and I. However, 47 by of axon A were missing at the C, A junction and axons F, G, and H had been skipped entirely. The reading frame shifts resulting from such aberrant RNA
processing showed that it could not corrEaspond exactly to the factor VIII sequence. At the C, A junction a good consensus splice site was utilized rather than the authentic one. The different splicing of the S36 clone compared with the authentic factor VIII transcript may be because only a portion of the RNA

primary transcript was expressed in the cos cell construction.
Alternatively, cell type or species variability may account for this difference.

7. cDNA Cloning a. Identification of a cell line producing Factor YIII mRNA
To identify a source of RNA for the isolation of factor VIII
cDWA clones, polyadenylated RNA was isolated from numerous human cell lines and tissues and screened by tJorthern blot hybriuization with the 189 by StuI-Hinc II fragment from the exon A region of x120. Poly(A)+ RNA from the CH-2 human T-cell hybridoma exhibited a hybridizing RNA species. The size of the hybridizing RNA was estimated to be about 10 kb. This is the size mRNA expected to code for a protein of about 300 kD. By comparison with control DNA
dot-blot hybridizations (66), the amount of this RNA was determined to be 0.0001-0.001 percent of the total cellular poly(A)+ RNA in the CH-2 cell line. This result indicated that isolation of factor VIII
cDNA sequences from this source would require further enrichment of specific sequences or otherwise entail the screening of extremely large numbers of cDNA clones.
b. Specifically Primed cDNA Clones The DNA sequence analysis of Factor VIIT genomic clones allowed the synthesis of 16 base synthetic oligonucleotides to specifically prime first strand synthesis of cDNA. Normally, oligo(dT) is used to prime cDNA synthesis at the poly(A) tails of mRNA. Specific priming has two advantages over oligo(dT). First, it serves to enrich the cDNA clone population for factor VIII. Second, it positions the cDNA clones in regions of the gene for which we Possessed hybridization probes, This is especially important in cloning such a large gene. As cDJJA clones are rarely longer than 1000-2000 base pairs, oligo(dT) primed clones would usu ally be undetectable with a probe prepared from most regions of the factor VIII gene. The strategy employed was to use DPJA fragments and sequence information from the initial exon A region to obtain specifically primed cDNA clones. We proceeded by obtaining a set of overlapping cDtJA clones in the 5' direction based upon the characterization of the earlier generation of cDNA clones. In order to derive the more 3' region of cDNA, we employed cDNA and genomic clone fragments from 3' axons to detect oiigo{dT) primed cDNA
clones. Several types of cDtdA cloning procedures were used in the course of this endeavor and will be described below.
The initial specific: cDNA primer, 5'-CAGGTCAACATCAGAG {"primer 1"; see Fig. 9) was synthesized as the reverse complement of the 16 3'-terminal residues of the axon A sequence. C-tailed cDNA was synthesized from 5 ug of CH-2 cell poly(A)+ RNA with primer 1, and annealed into G-tailed pBR322 as described generally in (67).
Approximately 100,000 resulting E. coli transformants were plated on 100 150 mm dishes and screened by hybridization (48) with the 189 by StuI/HincII fragment from the axon A region of the genomic clone x120 (Figure 4). One bona fide hybridizing clone ("p1.11") was recovered (see Fig. 9). DNA sequence analysis of pl.ll demonstrated identity with our factor VIiI genomic clones. The 447 by cDNA
insert in pl.ll contained the first 104 b of genomic axon A (second strand synthesis apparently did not extend back to the primer) and continued further into what we would later show to be axons 8 and C. The 5' point of divergence with axon A sequence was bordered by a typical RNA splice acceptor site (61).
Although the feasibility of obtaining factor VIII cDNA clones from the CH-2 cell line had now been demonstrated, further refinements were made. Efforts of several types were made to further enrich CH-2 RNA for factor VIII message. A successful strategy was to combine specifically primed first strand cDNA
synthesis with hybrid selection of the resulting single stranded cDNA. Primer 1 was used with 200 ug of poly(A)+ CH-2 RNA to synthesize single stranded cDNA. Instead of using DNA polymerase to immediately convert this to double stranded DNA, the single stranded DNA was hybridized to 2 ug of 189 by StuI/HincII genomic fragment DNA which had been immobilized on activated ABM cellulose paper (Schleicher and Schuell "Transa-Bind"; see (48). Although RNA is -51- 1 ~ 4 1 3 9 5 usually subject to hybrid selection, the procedure was applied after cDNA synthesis in order to avoid additional manipulation of the rare, large and relatively labile factor VIII RNA molecules. After elution, the material was converted to double stranded cDNA, size selected, and 0.5 ng of recovered DNA vas C-tailed and cloned into pBR322 as before. Approximately 12,000 recombinant clones were obtained and screened by hybridization with a 364 by Sau3A/StuI
fragment derived from the previous cDNA clone pl.ll. The probe fragment was chosen deliberately not to overlap with the DNA used for hybrid selection. Thus avoided was the identification of spurious recombinants containing some of the StuI/Ninc_II DNA
fragment which is invariably released from the DBM cellulose. 29 hybridizing colonies were obtained. This represents a roughly 250-fold enrichment of desired clones over the previous procedure.
Each of the 29 new recombinants was characterized by restriction mapping and the two longest (p3.12 and p3.48; Fig. 9) were sequenced. These cDNA clones extended about 1500 by farther 5' than pl.ll. Concurrent mapping and sequence analysis of cDNA and genomic clones revealed the presence of an unusually large exon (exon B, Fig. 4) which encompassed p3.12 and p3.48. Based on this observation, DNA sequence analysis of the genomic clone x222 was extended to define the extent of this exon. Exon B region contained an open reading frame of about 3 kb. 16 mer primers 2 and 3 were synthesized to match sequence within this large exon in the hope of obtaining a considerable extension in cDNA cloning.
At this point, it was demonstrated that a bacteriophage based cDNA cloning system could be employed, enabling production and screening of vast numbers of cDNA clones vrithout prior enrichment by hybrid selection. aGTlO (68) is a phage a derivative with a single EcoRI restriction site in its repressor gene. If double stranded cDPJA fragments are flanked by EcoRI sites they can be iigated into this unique site. Insertion of foreign DNA into this site renders the phage repressor minus, forming a clear plaque. aGTlO without insert forms turbid plaques which are thus distinguishable from recombinants. In addition to the great transformation efficiency ~ 341395 -52w inherent in phage packaging, a cDNA plaques are more convenient to screen at high density than are bacterial colonies.
Double stranded cDNA was prepared as before using primer 3, 5'-AACTCTGTTGCTGCAG (located about 550 by downstream from the postulated 5' end of exon B). EcoRI "adaptors" were ligated to the blunt ended cDNA. The adaptors consisted of a complementary synthetic l8mer and 22mer of sequence 5"-CCTTGACCGTAAGACATG and 5'-AATTCATGTCTTACGGTCAAGG. The 5' end of the l8mer was phosphorylated, while the 5' end of the 22mer retained the 5'-ON
with which it was synthesized. Thus, when annealed and ligated with the cDNA, the adaptors form overhanging FcoRI sites which cannot self-ligate. This allows one to avoid FcoRI methylation of cDNA and subsequent EcoRI digestion which follows linker ligation in other published procedures (83). After gel isolation to size select the cDNA and remove unreacted adapters, an equimolar amount of this cDNA
was ligated into EcoRI cut aGTlO, packaged and plated on _E. coli c600hf1~. About 3,000,000 clones from 1 ug of poly(A)+ RNA were plated on 50 150mm petri dishes and hybridization screened with a 300 by HinfI fragment from the 5' end of exon B. 46 duplicate 2U positives were identified and analyzed by EcoRI digestion. Several cDNA inserts appeared to extend about 2500 by 5'of primer 3. These long clones were analyzed by DNA sequencing. The sequences of the 5' ends of x13.2 and x13.27 are shown in Fig. 10. They possessed several features which indicated that Ne had reached the 5' end of the coding region for factor VIII. The initial 109 by contained stop codons in all possible reading frames. Then appeared an ATG
triplet followed by an open reading frame for the rest of the 2724 by of the cDNA insert in x13.2. Translation of the sequence following the initiator ATG gives a 19 amino acid sequence typical of a secreted protein "leader" or "pre" sequence (69). Its salient features are two charged residues bordering a 10 amino acid hydrophobic core. Following this putative leader sequence is a region corresponding to amino terminal residues obtained from protein sequence analysis of 210 kD and 95 kD thrombin digest species of factor VIII.

c. Oligo(dT) primed cDPjA clones Several thousand more 3' bases of factor VIII mRNA remained to be converted into cDNA. The choice was to prime reverse transcription with oligo(dT) and search for cDNA clones containing the 3' poly(A)+ tails of mRNA. However, in an effort to enrich the clones and to increase the efficiency of second strand DNA
synthesis, established procedures were replaced with employment of a specific primer of second strard cDNA synthesis. The 16-mer primer 4, 5'-TATTGCTGCAGTGGAG, was synthesized to represent message sense sequence at a PstI site about 400 by upstream of the 3' end of exon A (Fig. 9). mRNA vas reverse transcribed with oligo(dT) priming, primer 4 was added with DNA polymerase for second strand synthesis, and EcoRI adapted cDNA then ligated into aGTlO as before. 3,000,000 plaques were screened with a 419 by Pstl/HincII fragment contained on p3.12, lying downstream from primer 4. DNA was prepared from the four clones recovered. These were digested, mapped, and blot - _ hybridized with further downstream genomic fragments which had just been identified as exons using SV40 exon expression plasmids described above. Three of the four recombinants hybridized. The longest, x10.44, was approximately 1,800 base pairs. The DNA
sequence of x10.44 showed that indeed second strand synthesis began at primer 4. It contained all exon sequences found in the SV40 exon expression clone S36 and more. However, the open reading frame of x10.44 continued to the end of the cDNA. No 3' untranslated region nor poly(A) tail was found. Presumably second strand synthesis had not gone to completion.
To find clones containing the complete 3' end, we rescreened the same filters with labeled DNA from x10.44. 24 additional clones were recovered and mapped, and the two longest (x10.3 and x10.9.2) were sequenced. They contained essentially identical sequences which overlapped x10.44 and added about 1900 more 3' base pairs. 51 base pairs beyond the end of the x10.44 terminus, the DNA sequence showed a TGA translation stop codon followed by an apparent 3' untranslated region of 1805 base pairs. Diagnostic features of this region are stop codons dispersed in all three reading frames and a -54- ~ 3 4 ~ 3 9 5 poly(A) signal sequence, AATAAA (89),followed 15 bases downstream with a poly(A) stretch at the end of the cDNA (clone x10.3 contains 8 A's followed by the EcoRI adapter at this point, while x10.9.2 contains over 100 A's at its 3' end).
d. Complete cDNA Sequence The complete sequence of overlapping clones is presented in Figure 10. It consists of a continuous open reading frame coding for 2351 amino acids. Assuming a putative terminal signal peptide of 19 amino acids, the "mature" protein would therefore have 2332 amino acids. The calculated molecular weight for this protein is about 267,000 daltons. Taking into account possible glycosylation, this approximates the molecular weight of native protein as determined by SDS polyacrylamide gel electrophoresis.
The "complete" cDNA length of about 9000 base pairs (depending on the length of 3' poly(A)) agrees with the estimated length of the mRNA determined by Northern blot hybridization. The 5' (amino terminal coding) region contains substantial correspondence to the peptide sequence of 210 kD derived factor VIII material and the 3' 20 (carboxy terminal coding) region contains substantial correspondence to the peptide sequence of 80 kD protein.
8. Expression of Recombinant Factor VIII
a. Assembly of full length clone 25 In order to express recombinant Factor VIII, the full 7 kb protein coding region was assembled from several separate cDNA and genomic clones. We describe below and in Figure 11 the construction of three intermediate plasmids containing the 5', middle, and 3' regions of the gene. The intermediates are combined in an 30 expression piasmid following an SV40 early promoter. This plasmid in turn serves as the starting point for various constructions with modified terminal sequences and different promoters and selectable markers for transformation of a number of mammalian cell types.
The 5' coding region was assembled in a pBR322 derivative in 35 such a way as to place a CIaI restriction site before the ATG start codon of the Factor VIII signal sequence. Since no other CIaI site is found in the gene, it becomes a convenient site for refinements of the expression plasmid. The convenient CIaI and SacI containing plasmid pT24-10 (67a) was cleaved with HindIII, filled in with DNA
polymerase, and cut with SacI. A 77 b AIuI/SacI was recovered from the 5' region of the Factor YIII cDNA clone x13.2 and ligated into this vector to produce the intermediate called pF8Cla-Sac. (The AIuI site is located in the 5' unt:ranslated region of Factor VIII
and the SacI site 10 b beyond the initiator ATG at nucleotide position 10 in Fig. 10; the nucleotide position of all restriction sites to follow will be numbered as in Fig. 10 beginning with the A
of the initiator codon ATG.) An 85 b CIaI/SacI fragment containing 11 by of adaptor sequence (the adaptor sequence 5' ATCGATAAGCT is entirely derived from pBR322) was isolated from pFBCIa-Sac and ligated along with an 1801 b SacI/K~nI (nucleotide 1811) fragment from X13.2 into a CIaI/KpnI vector prepared from a pBR322 subclone containing a HindIII fragment (nuc. 1019-2277) of Factor VIII. This intermediate, called pF8Cla-Kpn, contained the initial 2277 coding nucleotides of Factor VIII preceded by 65 5' untranslated base pairs and the 11 base pair CIaI adaptor sequence. pF8Cla-Kpn was opened with K~nI and SphI (in the pBR322 portion) to serve as the vector fragment in a ligation with a 466 b K~nI/HindIII fragment derived from an EcoRI subclone of x13.2 and a 1654 b HindIII/S~hI (nuc.
4003) fragment derived from tire exon B containing subclone p222.8.
This produced pF8Cla-Sph containing the first 3931 b of Factor VIII
coding sequence.
The middle part of the coding region was derived from a three-piece ligation combining fragments of three pBR322/cDNA clones or subclones. p3.48 was opened with BamHI (nuc. 4743) and SaII (in pgR322 tet region) to serve as vector. Into these sites were ligated a 778 b BamHI/NdeI (nuc. 5520) fragment from p3.12 and a 2106 b NdeI/SaIT (in pBR322) fragment from the subclone pa10.44R1.9. Proper ligation resulted in a tetracycline resistant plasmid pFBSca-RI.
The most 3' portion of Factor VIII cDNA was cloned directly into _~6_ 1 3 4 1 ~ 9 5 an SY40 expression vector. The plasmid pCVSVEHBY contains an SY40 early promoter followed by a polylinker and the gene for the Hepatitis B surface antigen.
[pCYSVEHBV, also referred to as pCYSVEHBS, is a slight variant of p342E (73). In particular, pCYSVEHBV was obtained as follows:
The 540 by HindIII-HindIII fragment encompassing the SY40 origin of replication (74) was ligated into plasmid pMl (75) between the EcoRI
site and the HindIII site. The plasmid EcoRI site and SV40 HindIII
site were made blunt by the addition of Klenow DNA polymerise I in the presence of the 4 dNTPs prior to digestion with HindIII. The resulting plasmid, pESV, was digested with HindIII and BamHI and the 2900 b vector fragment isolated. Ti this fragment was ligated a HindIII-Bc~III fragment of 2025 b from HBV modified to contain a polylinker (DNA fragment containing multiple restriction sites) at the EcoRI site. The HBV fragment encompasses the surface antigen gene and is derived by EcoRI-aBgIII digestion of cloned HBY DNA
(74). The double stranded linker DNA fragment (5'dAAGCTTATCGATTCTAGAATTC3'...) was digested with HindIII and EcoRI
and added to the HBV fragment, converting the EcoRI->~III fragment to a-HindIII-B~III fragment. Although this could be done as a 3 part ligation consisting of linker, HBV fragment, and vector, it is more convenient and was so performed to first add the HindIII-EcoRI
linker to the cloned HBY DNA and then excise the HindIII-Bc~III
fragment by codigestion of the plasmid with those enzymes. The resulting plasmid, pCVSVEHBY, contains i bacterial origin of replication from the pBR322 derived pML, and ampicillin resistance marker, also from pML, an SV40 fragment oriented such that the early promoter will direct the transcription of the inserted HBV fragment, and the surface antigen gene from HBV. The HBV fragment also provides a polyadenylation signal for the productiors of polyadenylated mRNAs such as are normally formed in the cytoplasm of mammalian cells.
The plasmid pCVSVEHBV contained a useful CIaI site immediately 5' to an XbaI site in the polylinker. This plasmid was opened with XbaI and BamHI (in the Hepatitis Ag 3' untranslated region) and the _ 1 341 ~9 5 ends were filled in with DNA polymerase. This removed the Hepatitis surface antigen coding region but retained its 3' polyadenylation signal region, as well as the SV40 promoter. Into this vector was ligated a 1883 b EcoRI fragment (with filled in ends) from the cDNA
clone x10.3. This contained the final 77 coding base pairs of Factor VIII, the 1805 b 3' untranslated region, 8 adenosine residues, and the filled in EcoRI adaptor. By virtue of joining the filled in restriction sites, the EcoRI end was recreated at the 5' end (from filled in XbaI joined to filled in EcoRI) but destroyed at the 3' end (filled in EcoRI joined to filled in t3amHI). This piasmid was called pCVSVE/10.3.
The complete factor VIII cDt~JA region was joined in a three-piece ligation. pCYSVE/10.3 was opened with CIaI and EcoRI and served as vector for the insertion of the 3870 b CIaI/ScaI fragment from pF8Cla-Sca and the 3182 b Scal/EcoRI fragment from pFB.sca-RI. This expression plasmid was called pSYEFVIII.
b. Construction for Expression of Factor VIII in Tissue Culture Cells A variant vector based on PSYEFVIII, containing the adenovirus major late promoter, tripartide leader sequence, and a shortened Factor VIII 3'-untranslated region produced active factor VIII when stably transfected into BHK cells.
Figure 12 shows the construction of pAMt_3P.8c1, the expression plasmid that produces active factor VIII. To make this construction first the SstII site in pFDll (49r) and the ClaI site in pEHED22 (49y) were removed with Klenow DtdA polymerase I. These sites are in the 3' and 5' untranslated regions of the DHFR gene on these plasmids. Then a three-part ligation of fragments containing the deleted sites and the hepatitis B surface antigen gene from pCVSVEHBS (supra) was performed to generate the vector pCVSVEHE022eCS which has only one CIaI and one SstII site. The plasmid pSVEFYIII containing the assembled factor VIII gene (Figure 11) was cleaved with CIaI and HpaI to excise the entire coding region and about 380 b of the 3' untranslated region. This ~ 34~ 39 5 -58_ was inserted into the CIaI, SstII deletion vector at its unique CIaI
and H~aI sites, replacing the surface antigen gene to give the expression plasmid pSVE.8clD.
Separately, the adenovirus major late promoter with its tripartite 5' leader was assembled from two subclones of portions of the adenovirus genome along with a DHFR expression plasmid, pEND22 (49y). Construction of the two adenovirus subclones, pUCHSX and pMLP2 is described in the methods. pMLP2,contains the SstI to HindIII fragment from adenovirus coordinates 15.4 to 17.1 cloned in the SstI to HindIII site of pUCl3 (59). pUCHSX contains the HindIII
to XhoI fragment coordinates 17.1 to 26.5 cloned in the HindIII to SaII site of pUCl3. When assembled at tt~e HindIII site, these two adenovirus fragments contain the major late promoter of adenovirus, all of the first two exons and introns, and part of the third exon up to the XhoI site in the 5' untranslated region.
A three-part ligation assembled the adenovirus promoter in front of the DHFR gene in the plasmid pAP9L3P.D22.__ This put a CIaI site shortly following the former XhoI site in the third exon of the adenovirus tripartite 5' leader. Finally, the SV40 early promoter of the factor VIII expression plasmid, pSVE.BclD~was removed with CIaI and SaII and replaced with an SV40 early/adenovirus tandem promoter (see Figure 12) to generate the final expression plasmid, pAML3P.8c1. This plasmid contains the adenovirus tripartite leader spliced in the third exon to the 5' untranslated region of factor VIII. This is followed by the full length Factor VIII: structural gene including its signal sequence. The 3' untranslated region of the factor VIII gene i~; spliced at the H~aI site to the 3' untrans-lated region of Hepatitis B surface antigen gene. This is followed by the DHFR gene which has an SV40 early promoter and a Hepatitis 3' untranslated region conferring a functional polyadenylation signal.
The factor VIII expression plasmid, pAML3P.8c1, was cotransfected into BHK cells with the neomycin resistance vector pSVEneoBal6 (ATCC
tVo. CRL 8544, deposited 20 April 1984). These cells were first selected with 6418 followed by a selection with methotrexate.
Initial characterization of the Factor VIII RHA produced by the -59_ 1 3 41 3 9 5 BHK cell line was performed by Northern analysis of poly(A)+ cyto-plasmic RNA by hybridization to a 32P-labeled Factor VIII DNA probe.
This analysis shows a band approximately 9 kb in length. Based on hybridization intensities, this band is about 100 to 200 fold enriched when compared to the 9 kb band found in the CH-2 cell line.

9. Identification of Recombinant Factor VIII
a. Radioimmune assay Radioimmune assays were performed as described in the Methods on supernatants and lysed cells from the BHK Factor VIII producing cell line. Table 1 shows that the supernatants (which contain factor VIII
activity) (see 96) also contain approximately equal amounts of the 210 kD (C10) and the 80 kD (C7F7) portions of Factor VIII as judged by these RIAs. Factor VIII can also be detected in the cell lysates by both RIAs. Control cell lines not expressing factor VIII produced RIA values of less than 0.001 units per ml.
Table 1 Factor VIII RIA of BHK cell line transfected with pAML3P.8c1 Cell Supernatant Exp. 1 0.14 U/ml 0.077 Exp. 2 0.022 0.021 Cell Lysate Exp. 1 0.4. 0.016 1125 cpm bound were converted to units/ml with a standard curve based on dilutions of normal plasma. All values are significantly above barkgrour~d. Limits of detection were 0.005 U/ml for the CIO and 0.01 U/ml for the C7F7 assays.
b. Chromogenic Assay on BNK Cell Media As is shown in Table 2, media from these cells generated an absorbance at 405 nm when tested in the Coatest assay. As described above, this assay is specific for factor VIII activity in the activation _60- '1 ~ 4 ~ 3 9 5 of factor X. Addition of monoclonal antibodies specific for factor YIII decreased the amount of factor Xa generated as evidenced by the decrease in absorbance from 0.155 for the media to 0.03 for the media plus antibodies (after subtracting out the blank value).
Therefore, the cells are producing an activity which functions in an assay specific for factor VIII activity and this activity is neutralized by antibodies specific for factor VIII.
Incubation of the media in the reaction mixture without the addition of the factor lXa~ factor X, and phospholipid did not result in an increase in the absorbance at 405 nm above the blank value. The observed activity is therefore not due to the presence of a nonspecific protease cleaving the substrate, and in addition neutralized by antibodies specific for factor VIII.
Table 2 Factor VIII Activity of 8HK cell line determined by chromogenic assayl.
Absorbance Absorbance at 405 nrn at (Control value Sample 405 nm subtracted) Piedi a 0.193 0.155 Buffer control2 0.038 (0.0) Media + factor VIII antibody3 0.064 0.030 Buffer control2 + factor VIII antibody 0.034 (0.0) 1 Reactions modified as follows: 50 ui each of Txa/x/phospholipid, CaCl2~ and '!-100 diluted sample were incubated 10 minutes at 37'C. S2222 (50 u1 was added and reaction terminated with 100 ul~
34 of 50 percent acetic acid after 60 minutes at 37'C.
2 Buffer used in place of sample was 0.05 M Tris - HC1, pH 7.3, containing 0.2 percent bovine serum albumin.
3 Antibody was a mixture of C8 and C7F7 (10 ug each). The media was preincubated 5 minutes prior to start of assay.

_61- 1 341 39 5 C . Chrr~rnatnrrra~y pf i a on Mon ~1 onal R S i n Serum containing media containing factor VIII activity was chromatographed on the C8 monoclonal antibody (ATCC No. 40115, deposited 20 April 1984) column as described (supra). The eluted fractions were diluted 1:100 and assayed for activity. To 50 u1 of the diluted peak fraction was added various monoclonal antibodies known to be neutralizing for plasma factor VIII activity. The results shown in Table 3 demonstrate that the factor VIII activity eluted from the column (now much more concentrated than the media) w'as also neutralized by these factor VIII antibodies.
Table 3 Chromogenic Assayl of Peak Fraction of Monoclonal Antibody Eluate Absorbance at _ Sampl a 405 nMl Peak Fraction2 0.186 Peak fraction plus factor VIII Antibody3 0.060 Buffer Control 0.000 Buffer Control plus factor VIII Antibody 0.045 1 Assay was performed as follows: 50 u1 of diluted sample was incubated 5 minutes with 50 w1 of Ik~/x/phosphol.ipid solution at 37'C. The reaction was incubated with 50 u1 of CaCl2~ and allowed to proceed 10 minutes at 37'C. The chromogenic substrate (50 u1) was added, and the reaction terminated by the addition of 100 u1 of 50 percent acetic acid after 10 minutes.
2 Peak fraction was diluted 1:100 in 0.05 t~1 Tris, pH 7.2, containing 0.15 !d NaCI for assays.
Antibody was 10 u1 of Symbiotic antibody added to the diluted sample and incubated 5 minutes at roam temperature.
d. Coagulant Activity of Purified Factor VIII.
The activity detected in the cell media was purified and concentrated by passage over a C8 monoclonal resin (supra). The Peak fraction was dialyzed against 0.05 M imidazole, pH6.9, containing 0.15 M NaCI, 0.02 M glycine ethyl ester, 0.01 M CaCl2~
and 10 percent glycerol in order to remove the elution buffer. The activity peak fraction was assayed by coagulation analysis in factor VIII deficient plasma (Table 4). A fibrin clot was observed at 84 seconds. With no addition, the hemophilia plasma formed a clot in 104.0 seconds. Therefore, the eluted fraction corrected the coagulation defect in hemophilia plasma. Normal human plasma was diluted and assayed in the same manner. A standard curve prepared from this plasma indicated that the eluted fraction had approxi-mately 0.01 units per milliliter of factor VITI coagulant activity.
Table 4 Coagulant Activity of Monoclonal Antibody Purified Factor VIII
Sample Clotting time (sec.) Recombinant factor VIII la 86.5 Recombinant factor VIII la and C7F7 antibody 101.3 Control2 _ 101.3 Recombinant factor VIII 1b 82.4 Recombinant factor VIII 1b and 10a Synbiotic antibody 110.6 Control2 95.5 Factor VIII was peak fraction eluted from the C8 monoclonal resin and dialyzed for 1 1/2 (la) or 2 (1b) hours in order to 2 remove elution buffer.
Control buffer was 0.05 M Tris, pH7.3, containing 0.2 percent bovine serum albumin.
e. Thrombin Activation of Purified Factor VIII.
Activation of coagulant activity by thrombin is a well established property of factor VIII. The eluted fraction from the monoclonal column was analyzed for this property. After dialysis of the sample to remove the elution buffer (supra), 100 ui of the eluate was diluted with 100 ~1 of 0.05 M imidazole, pH7.6, ._ 1 .34139a containing 0.15 M NaCI, 0.02 M glycine ethyl ester, 0.01 M CaCl2 and 10 percent glycerol. This dilution was performed to dilute further any remaining elution buffer (which might interfere with thrombin functioning) as well as to increase the pH of the reaction mixture. Thrombin (25 ng) was added to the solution and the reaction was performed at roorrr temperature. Aliquots of 25 u1 were removed at various time points, diluted 1:3, and assayed for coagulation activity. The results are shown in Figure 17. The factor VIII activity increaser.( with time, and subsequently decreased, as expected for a fiactor VIII activity. The amount of thrombin added did not clot factor VIII deficient plasma in times observed for these assays, and the observed time dependent increase and subsequent decrease in observed coagulation time proved that the activity being monitored was in fact due to thrombin acaivation of factor VIII. The observed approximately 20-fold activation by thrombin is in agreement with that observed for plasma factor VIII.
f. _ Binding of Recombinant Factor VIII to Immobilized von Willebrand Sep harose.
Factor VIII is knovrn to circulate in plasma in a reversible complex with von Willebrand Factor (vWF) (10-20). A useful form of recombinant factor VIII should therefore also possess this capacity for forming such a complex in order to confirm identity as factor VIII. In addition, the ability to form such a complex would prove the ability of a recombinant factor VIII to form the n<~tural, circulating form of the activity as the factor VIII/vWf complex upon infusion into hemophiliacs. In order to test the ability of recombinant factor VIII to interact with vWF, vWF was purified and immobilized on a resin as follows:
Human von Willebrand factor was prepared by chromatography of human factor VIII concentrates (purchased from, e.g., Cutter Laboratories) on a Sepharose CL4B resin equilibrated with 0.05 M
Tris, pH 7.3, containing 0.15 M ~JaCI. The von Willebrand factor elutes at the void volume of the column. This region was pooled, concentrated by precipitation with ammonium sulfate at 40 percent of saturation and re-chromatographed on the column in the presence of .. , _~4_ 1 3 41 3 9 5 the above buffer containing 0.25 M CaCl2 in order to separate the factor VIII coagulant activity from the von Willebrand factor. The void volume fractions were again pooled, concentrated using ammonium sulfate, and dialyzed against 0.1 M sodium bicarbonate. The resulting preparation was covalentiy attached to cyanogen bromide activated Sepharose (purchased from Pharmacia) as recommended by the manufacturer. The column was washed with C).~J2 M TriS, pH 7.3, containing 0.05 M NaCI and 0.25 t~l CaCl2 in order to remove unbound proteins.
The recombinant factor VIII was prepared in serum free media and applied to a 1.0 ml column of the vWF resin at room temperature.
The column was washed to remove unbound protein and eluted with 0.02 M Tris, pH 7.3, containing 0.05 M NaCi and 0.25 t4 CaCl2.
Fractions of 1.0 ml were collected, diluted 1:10 and <issayed. The results are shown in Table 5. The factor VIII activity is absorbed from the media onto the column. The activity can subsequently be eluted from the column using high salt (Table 5), as expected for the human factor VIII. Therefore, the factor VIII produced by the BHK cells has the property of specific interaction with the von Willebrand factor protein.
Table 5 Absorbance Sarn 1e at 405 nml Cell Media 0.143 2b Wash 0.015 Eluted Fractions 1 o.oaa 2 0.410 3 0.093 4 0.017 3a 5 0.000 6 0.000 Assay procedures were that recommended by the manufacturer, 35 except that all volumes were decreased by one half.

-65- 'I 3 41 3 9 5 10. Analysis of Fusion Proteins The purpose of this set of experiments was to prove immunological identity of the protein encoded by the clone with the polypeptides in plasma. This was accomplished by expressing portions of the gene as fusion proteins in E. coli. All or part of the coding sequences of the cloned gene can be expressed in forms designed to provide material suitable for raising ant;ibodies. These antibodies, specific for desired regions of the cloned protein, can be of use in analysis and purification of proteins. A series of E~ coli/factor VIII "fusion proteins" were prepared for this purpose. Fragments of factor VIII clones were ligated into the B~C1II site of the plasmid pNCY (70) in such a way as to join factor VIII coding sequences, in proper reading frame, to the first 12 amino acids of the fused E. coli trp LE protein (48, 70, 71).
Substantial amounts of recombinant protein product are usually produced from this strong try promoter system.
pfusl was constructed by isolating a 189 by StuI/HincII fragment of factor VIII (coding for amino acids 1799-1860) and ligating this into the SmaI site of pUCl3 (49K). This intermediate plasmid was digested with BamHI and EcoRI and the 200 by fragment inserted into pNCV (70) from which the 526 by III to EcoRI fragment had been removed. This plasmid, pfusl, produces under trp promoter control a 10 kD fusion protein consisting of i6 tar LE and linker coded amino acids, followed by 61 residues of factor VIII and a final 9 linker coded and trpE carboxy terminal residues.
pfus3 was constructed by removing a 290 by AvaII fragment of factor VIII (amino acids 1000-1096), filling in the overhanging nucleotides using Klenow fragment of DNA polymerase, and ligating this now blunt-ended DNA fragment into pNCV which had been cut with Bc~II and similarly filled in. This plasmid, with the filled in fragment in the proper orientation (as determined by restriction digests and DHA sequence analysis), directs the synthesis of an approximately 40 kD fusion protein containing 97 amino acids of factor VIII embedded within the 192 amino acid tr~LE protein.
pfus4 was made by cutting a factor VIII subclone, a222.8~with -~~- ~ 34~ 39 5 BanI, digesting back the overhang with nuclease S1, followed by PstI digestion and isolation of the resulting 525 by blunt/PstI
fragment (amino acids 710-885). This was ligated into pNCY, which had been digested with B~III, treated with S1, digested with PstI, and the vector fragment isolated. pfus4 directs the synthesis of a 22 kD fusion protein containing 175 amino acids of factor VIII
following the initial 12 amino acids of traLE.
The fusion proteins ~aere purified and injected into rabbits in order to generate antibodies as described Supra. roese antibodies were tested for binding to plasma derived factor VIII by Western Slot analysis.
The results of such a Western transfer are shown in Figure 13.
Each of the fusion proteins reacts with the plasma factor VIII.
Fusion 1 was generated from the region of the gene encoding an ~5 80,000 dalton polypeptide. It can be seen that fusion 1 antisera react only with the 8p,000 dalton Land, and do nat react with the proteins of higher molecular weight. Fusion 3 and 4 antisera show cross reactivity with the proteins of greater than 80,000 daltons, and do not react with the 80,000 dalton band. The monoclonal 20 antibody C8 is an activity neutralizing monoclonal directed against factor VIII and is known to react with the 210,000 dalton protein.
Figure 14 demonstrates that fusion 4 protein will react with this monoclonal antibody, thereby demonstrating that the amino acid sequence recognized by C8 is encoded by fusion 4 polypeptide. This 25 further supports the identity of fusion 4 protein containing protein sequences encoding the 210,000 dalton protein. The above studies conclusively prove that the gene encodes the amino acid sequence for both the 210,000 and 80,000 dalton proteins.
30 11. Pharmaceutical Compositions The compounds of the present invention can be formulated according to known methods to prepare pharmaceutically useful compositions, whereby the human factor VIII product hereof is combined in admixture with a pharmaceutically acceptable carrier 35 vehicle. Suitable vehicles and their formulation, inclusive of , . _57_ other human proteins, e.g. human serum albumin, are described for example in Remington's Pharmaceutical Sciences by E.W. Martin., which is hereby incorporated by reference. Such compositions will contain an effective amount of the protein hereof together with a suitable amount of vehicle in order to prepare pharmaceutically acceptable compositions suitable for effective administration to the host. For example, the human factor VIII hereof may be parenterally administered to subjects suffering, e.g., from hemophilia A.
The average current dosage for the treatment of a hemophiliac varies with the severity of the bleeding episode. The average doses administered intraveneously are in the rwnge of: 40 units per kilogram for pre-operative indications, 15 to 20 units per kilogram for minor hemorrhaging, and 20 to 40 units per kilogram administered over an 8 hour period for a maintenance dose.

a _68- 1 3 4 1 3 9 5 Bibliography 1. Hemostasis and Thrombosis, Bloom and Thomas, Eds., Churchill, Livingstone, NY (1981).
2. Davie, et al., Ann. Rev. Biochem. 44, 799 (1975).
3. Jackson, et al., Ann. Rev. Biochem. 49, 767 (1980).
4. Wright, J. Am. Mec. Assoc. 170, 32~> (1959).
5. Schmer, et al., J. Biol. Chem. 247, 2512 (1972).
6. Legaz, et al., J. Biol. Chem. 247, 3946 (1973).
7. Shapiro, et al., J. Clin. Invest. 52, 2198 (1973).
8. Nilsson, et al., Acta. Med Scanl. 159, 179 (1957).
9. McKee, Ann. NY Acad. Sci. 240, 8 (1975).
Arch. Biochem. Bio h s 95 70 (1961).
10. Theiin, et al., p y .

11. Griggs, et ai., Proc. Natl. Acad. Sci. 70, 2814 (1973).

12. Donati, et al., Thromb. Res. 2, 97 (1973).

13. Weiss, et al., Br. J. Haematol 18, 89 (1970).
- -14. Weirs, et al., Thromb. Diath. Haemorrh 27, 212 (1972).

15. Weiss, et al., Science 182, 1149 (1973).

16. Rick, et al., Blood 42, 737 (1973).

17. Rick, et al., Thromb. Res. 7, 909 (1975).

18. Thelin, et al., Arch. Biochem. Biophys. 95, 70 {1961).

19. Owen, et al., Thromb. Diath. Haemorrh. 27, 502 (1972).

20. Cooper, et al., Proc. Natl. Acad. Sci. 70, 2326 {1.973).

21. Vehar, et al., Biochemistry 19, 401 (1980).

22. Fulcher, et al., Proc. Hatl. Acad. Sci. 79, 1648 (1982).

23. Fulcher, et al., Blood 61, 807 (1983).

24. Fulcher, et al., Blood 63, 486 (1984).

-6~~ 9 3 41 3 9 5 Bibliography Continued 25, Fass, et al., Blood 59, 594 (1982). _ 26. Knutson, et al., Blood 59, 615 (1982).

27. Fay, et al., Proc. Natl, Acad. Sci. 79, 7200 (1982).

28. Gordon, et al., Thrombosis Res. 6, 127 (1975).

29. Sacharski, et aT., Mayo Clinic Proc. 44, 784 (1969).

30. Green, et al., Brood 37, 47 (1971}.
~0 31. Schwinn, et al., U.S. Patent 4067964.
31a. Zimmerman, et al., Haemostasis and Thrombosis, Bloom and Thomas, Eds., Churchill, Livingstone, NY, p.lll (1981).
32. Bolivar, et al., Gene 2, 95 (1977).
33. Chang, et al., Nature 275, 615 (1978).
_ 34. Itakura, et al., Science 198, 1056 (1977).
35. Goeddel, et al., Nature 281, 544 (1979).
-36. Goeddel, et al,, tJucleic Acids Res. 8, 4057 (1980).
37. European Patent Appln. Publ. No. 0036776.
38. Siebenlist, et al., Cell 20, 269 (1980).
39. Stinchcomb, et al., Nature 282, 39 (1979).
40. Kingsman, et al., Gene 7, 141 (1979).
41. Tschemper, et al., Gen~._ 10, 157 (I980), 42. Jones, Genetics 85, i2 (1977), 43. Hitzeman, et aT., J. Biol. Chem. 255, 2073 (1980).
44. Hess, et al., J~ Adv. Enzyme Reg. 7, 149 (1968).
45. Holland, et al., Biochemistry I7, 4900 (1978).
46. Graham, et al., Vir-_ ology 52, 546 (I978).
47. Cohen, et al., PIJAS (USA) 69, 2110 (1972).
-- - -..

°

- ~ 34~ 3g ~
Bibliography Continued 47a. Crea, et al., Nucleic Acids Res. 8, 2331 (1980) 47b. Beaucage, et al., Tetrahedron Lett. 22, 1859 (1981) 48. Maniatis, et al., P4oiecuiar Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Narbor, N.Y. (1982).
48a. Lawn, et al., Nucleic Acids Res. 9, 1045 (1981).
49. Yarn, et al., Proc. Natl. Acad. Sci. USA 77, 5172 (1980).
49a. Southern, J. Molec. Biol. 98, 503 (1975).
49b. Goidberg, Proc. Natl. Acad. Sci. USA 77, 5794 (1980).
49c. Taylor, et al., Biochem. biophys. Acta 442, 324 (1976).
49f. Ullrich, et al., Science 196, 1313 (1977).
49g. Berger, et al., Biochemistry 18, 5143 (1979).
49h. Aviv, et al., Proc. Natl. Acad. Sci. USA 69, 1408 (1972).
49i. Sanger, et al., Proc. Natl. Acad. Sci. USA 74, 5463 (1977).
49~~ Gingeras, et al., J. Biol. Chem. 257 13475 (1982).
49k. Norrander, et al., Gene 26, 101 (1983).
491. Picken, et al., Inf, and Immun. 42, 269 (1983).
49m. Beck, et al., Gene 19, 327 (1982).
49n. Simonson, et al., Mol. Cell Biol. 3, 2250 (1983).
490. Southern, et al., J, h1ol. and Applied Gen. 1, 327 (1982).
49p. Graham, et al., Virology 52, 456 (1978).
49q. Wigler, et al., Proc. Natl. Acad. Sci. USA 77, 3567 (1980).
49r. Simonson, et al., Proc. Natl. Acad. Sci. USA 80, 2495 (1983).
49s. Rotblat, et al., Thromb. Res. 21, 431 (1981).
49t. Rotblat, et al., J. Lab. Clin. Med. 101, 736 (1983).
49u. Kearney, et al., J. Immun. 123, 1548 (1979).

a Bibliography Continued 49v. Kohler ,et al., Nature 256, 495 11975).
49w. Benton, et al., Yirol. 117, 490 (1982).
49x. Chaion, et al., J. Immun. 9, 359 (1979).
49y. U.S.S.N. 459152, filed 19 Jan. 83.
50. Ish-Horowitz, et al., Nucl. Acids Res. 9, 2989 (1981.).
51. Stel, et al., PJature 303, 530 (1983).
52. Kelly, et al., Brit. J. Haem. (1984).
53. Exmer, et al., Thrombosis Res. 32, 427 (1983).
54. Jaffe, et al., J. Clin. Invest. 53, 36a (1974).
55. Fay, et al., Proc. Natl. Acad. Sci. USA 79, 7200 (1982).
56. Zimmerman, et al,, Haemostasis and Thrombosis Bloom, A.L. and Ouncan, P.T. (Churchill Livinstone, New York) pp. 111-123 (1981).
56a. Meili, et al., Thromb. Oiathes. Haemorrh. 24, 161 (1970).
_.. -57. Grantham, et al., Nuclei Arts Res. 8, r49 (1980).
58. Kurachi, et al., Proc. natl. Acad. Sci. USA 79, 6461 (1982).
59. Viera, et al., Gene 19, 259 (1982).
60. Breathnach, et al.., Ann. Rev. Biochem 50, 349 (1981).
61. Sharp, Cell 23, 643 (1981).
62. Fritsch, et al., Cell 19, 959 (1980).
63. Collies, et al., Proc. Natl. Acad. 5ci. USA 75, 4242 (1978?.
66. Kafatos, et al., Nucleic Acids Res. 7, 1541 (1979).
67. Capon, et al., Nature 304, 507 (1983).
67a. Capon, et al,, Nature 301, 33 (1983).

-72- 1 3 41 3 9 ~
Bibliography Continued 68. Huynh, et al., Practical Approaches in Biochemistry (IRL Press Ltd., Oxford, England) (1984).
'' 69. Perlman, et al., J. Mol. Biol. 167, 391 (1983).

70. Kleid, et al., Science 214, 1125 (1981).

71. Goeddel, et al., Nature 287, 411 (1980).

73. Crowley, et al., Mol. Cell Biol. 3, 44 (1983).

74. Liu, et al., Dt~A 1, 213 (1982).

75. Lusky, et al., Nature 293, 79 (1981).

76. Hanahan, J. Mol. Biol. 166, 557 (1983).

77. Gluzman Cell 23, 175 (1981).

1~

78. Grunstein, et al., Proc. Natl. Acad. Sci. USA 72, 3961 (1975).

79. Tuddenham, et al., Symposium on Factors VIII/von Willebrand Factor, October 7-9, 1982, p.1.

2U 8U. Morrissey, Anal. Bioclem 117, 307 (1981).

81. Laemmli, Nature 227, 680 (1970), 82. Greenwood, et al., Biochem. J. 89, 114 (1963).

83. Maniatis, et al., Cell 15, 687 (1963).

84. Sompayrac. et al., Proc. Natl. Acad. Sci. USA 78, 7575 (1981).

86. h1essing, et al., Nucl. Acids Res. 9, 304 (1981).

87. Wood, et al., J. Biol. Chem. 256, 1502 (1981).

88. Shen, et al., Cell 19, 379 (1981).

30 89. Proudfoot, et al., Nature 252, 359 (1976).

Claims (3)

1. A recombinant human Factor VIII, having activity which corresponds to factor VIII activity native to human plasma, selected from the group consisting of a Factor VIII protein having the amino acid sequence of Figure 10 from amino acid residues S1. to 2332, a Factor VIII protein having the amino acid sequence of Figure 10 from amino acid residues 1 to 2332, a Factor VIII protein having the amino acid sequence of Figure 10 from amino acid residues S1 to 2332 except that the amino acid at position 1241 is glutamic acid encoded by GAG instead of aspartic acid encoded by GAC, and a Factor VIII protein having the amino acid sequence of Figure 10 from amino acid residues 1 to 2332 except that the amino acid at position 1241 is glutamic acid encoded by GAG instead of aspartic acid encoded by GAC, which Factor VIII is free of viral contaminants that infect humans.

2. The recombinant functional human Factor VIII of claim 1 produced by non-human cells.

3. A pharmaceutical composition comprising a recombinant human Factor VIII of any one of claims 1 to 2 together with a pharmaceutically acceptable carrier, which composition is free of viral contaminants that infect humans.